Practical Machine Learning with R and Python – Part 5


This is the 5th and probably penultimate part of my series on ‘Practical Machine Learning with R and Python’. The earlier parts of this series included

1. Practical Machine Learning with R and Python – Part 1 In this initial post, I touch upon univariate, multivariate, polynomial regression and KNN regression in R and Python
2.Practical Machine Learning with R and Python – Part 2 In this post, I discuss Logistic Regression, KNN classification and cross validation error for both LOOCV and K-Fold in both R and Python
3.Practical Machine Learning with R and Python – Part 3 This post covered ‘feature selection’ in Machine Learning. Specifically I touch best fit, forward fit, backward fit, ridge(L2 regularization) & lasso (L1 regularization). The post includes equivalent code in R and Python.
4.Practical Machine Learning with R and Python – Part 4 In this part I discussed SVMs, Decision Trees, validation, precision recall, and roc curves

This post ‘Practical Machine Learning with R and Python – Part 5’ discusses regression with B-splines, natural splines, smoothing splines, generalized additive models (GAMS), bagging, random forest and boosting

As with my previous posts in this series, this post is largely based on the following 2 MOOC courses

1. Statistical Learning, Prof Trevor Hastie & Prof Robert Tibesherani, Online Stanford
2. Applied Machine Learning in Python Prof Kevyn-Collin Thomson, University Of Michigan, Coursera

You can download this R Markdown file and associated data files from Github at MachineLearning-RandPython-Part5

The content of this post and much more is now available as a compact book  on Amazon in both formats – as Paperback ($9.99) and a Kindle version($6.99/Rs449/). see ‘Practical Machine Learning with R and Python – Machine Learning in stereo

For this part I have used the data sets from UCI Machine Learning repository(Communities and Crime and Auto MPG)

1. Splines

When performing regression (continuous or logistic) between a target variable and a feature (or a set of features), a single polynomial for the entire range of the data set usually does not perform a good fit.Rather we would need to provide we could fit
regression curves for different section of the data set.

There are several techniques which do this for e.g. piecewise-constant functions, piecewise-linear functions, piecewise-quadratic/cubic/4th order polynomial functions etc. One such set of functions are the cubic splines which fit cubic polynomials to successive sections of the dataset. The points where the cubic splines join, are called ‘knots’.

Since each section has a different cubic spline, there could be discontinuities (or breaks) at these knots. To prevent these discontinuities ‘natural splines’ and ‘smoothing splines’ ensure that the seperate cubic functions have 2nd order continuity at these knots with the adjacent splines. 2nd order continuity implies that the value, 1st order derivative and 2nd order derivative at these knots are equal.

A cubic spline with knots \alpha_{k} , k=1,2,3,..K is a piece-wise cubic polynomial with continuous derivative up to order 2 at each knot. We can write y_{i} = \beta_{0} +\beta_{1}b_{1}(x_{i}) +\beta_{2}b_{2}(x_{i}) + .. + \beta_{K+3}b_{K+3}(x_{i}) + \epsilon_{i}.
For each (x{i},y{i}), b_{i} are called ‘basis’ functions, where  b_{1}(x_{i})=x_{i}b_{2}(x_{i})=x_{i}^2, b_{3}(x_{i})=x_{i}^3, b_{k+3}(x_{i})=(x_{i} -\alpha_{k})^3 where k=1,2,3… K The 1st and 2nd derivatives of cubic splines are continuous at the knots. Hence splines provide a smooth continuous fit to the data by fitting different splines to different sections of the data

1.1a Fit a 4th degree polynomial – R code

In the code below a non-linear function (a 4th order polynomial) is used to fit the data. Usually when we fit a single polynomial to the entire data set the tails of the fit tend to vary a lot particularly if there are fewer points at the ends. Splines help in reducing this variation at the extremities

library(dplyr)
library(ggplot2)
source('RFunctions-1.R')
# Read the data
df=read.csv("auto_mpg.csv",stringsAsFactors = FALSE) # Data from UCI
df1 <- as.data.frame(sapply(df,as.numeric))
#Select specific columns
df2 <- df1 %>% dplyr::select(cylinder,displacement, horsepower,weight, acceleration, year,mpg)
auto <- df2[complete.cases(df2),]
# Fit a 4th degree polynomial
fit=lm(mpg~poly(horsepower,4),data=auto)
#Display a summary of fit
summary(fit)
## 
## Call:
## lm(formula = mpg ~ poly(horsepower, 4), data = auto)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -14.8820  -2.5802  -0.1682   2.2100  16.1434 
## 
## Coefficients:
##                       Estimate Std. Error t value Pr(>|t|)    
## (Intercept)            23.4459     0.2209 106.161   <2e-16 ***
## poly(horsepower, 4)1 -120.1377     4.3727 -27.475   <2e-16 ***
## poly(horsepower, 4)2   44.0895     4.3727  10.083   <2e-16 ***
## poly(horsepower, 4)3   -3.9488     4.3727  -0.903    0.367    
## poly(horsepower, 4)4   -5.1878     4.3727  -1.186    0.236    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 4.373 on 387 degrees of freedom
## Multiple R-squared:  0.6893, Adjusted R-squared:  0.6861 
## F-statistic: 214.7 on 4 and 387 DF,  p-value: < 2.2e-16
#Get the range of horsepower
hp <- range(auto$horsepower)
#Create a sequence to be used for plotting
hpGrid <- seq(hp[1],hp[2],by=10)
#Predict for these values of horsepower. Set Standard error as TRUE
pred=predict(fit,newdata=list(horsepower=hpGrid),se=TRUE)
#Compute bands on either side that is 2xSE
seBands=cbind(pred$fit+2*pred$se.fit,pred$fit-2*pred$se.fit)
#Plot the fit with Standard Error bands
plot(auto$horsepower,auto$mpg,xlim=hp,cex=.5,col="black",xlab="Horsepower",
     ylab="MPG", main="Polynomial of degree 4")
lines(hpGrid,pred$fit,lwd=2,col="blue")
matlines(hpGrid,seBands,lwd=2,col="blue",lty=3)

fig1-1

1.1b Fit a 4th degree polynomial – Python code

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import PolynomialFeatures
from sklearn.linear_model import LinearRegression
#Read the auto data
autoDF =pd.read_csv("auto_mpg.csv",encoding="ISO-8859-1")
# Select columns
autoDF1=autoDF[['mpg','cylinder','displacement','horsepower','weight','acceleration','year']]
# Convert all columns to numeric
autoDF2 = autoDF1.apply(pd.to_numeric, errors='coerce')

#Drop NAs
autoDF3=autoDF2.dropna()
autoDF3.shape
X=autoDF3[['horsepower']]
y=autoDF3['mpg']
#Create a polynomial of degree 4
poly = PolynomialFeatures(degree=4)
X_poly = poly.fit_transform(X)

# Fit a polynomial regression line
linreg = LinearRegression().fit(X_poly, y)
# Create a range of values
hpGrid = np.arange(np.min(X),np.max(X),10)
hp=hpGrid.reshape(-1,1)
# Transform to 4th degree
poly = PolynomialFeatures(degree=4)
hp_poly = poly.fit_transform(hp)

#Create a scatter plot
plt.scatter(X,y)
# Fit the prediction
ypred=linreg.predict(hp_poly)
plt.title("Poylnomial of degree 4")
fig2=plt.xlabel("Horsepower")
fig2=plt.ylabel("MPG")
# Draw the regression curve
plt.plot(hp,ypred,c="red")
plt.savefig('fig1.png', bbox_inches='tight')

fig1

1.1c Fit a B-Spline – R Code

In the code below a B- Spline is fit to data. The B-spline requires the manual selection of knots

#Splines
library(splines)
# Fit a B-spline to the data. Select knots at 60,75,100,150
fit=lm(mpg~bs(horsepower,df=6,knots=c(60,75,100,150)),data=auto)
# Use the fitted regresion to predict
pred=predict(fit,newdata=list(horsepower=hpGrid),se=T)
# Create a scatter plot
plot(auto$horsepower,auto$mpg,xlim=hp,cex=.5,col="black",xlab="Horsepower",
     ylab="MPG", main="B-Spline with 4 knots")
#Draw lines with 2 Standard Errors on either side
lines(hpGrid,pred$fit,lwd=2)
lines(hpGrid,pred$fit+2*pred$se,lty="dashed")
lines(hpGrid,pred$fit-2*pred$se,lty="dashed")
abline(v=c(60,75,100,150),lty=2,col="darkgreen")

fig2-1

1.1d Fit a Natural Spline – R Code

Here a ‘Natural Spline’ is used to fit .The Natural Spline extrapolates beyond the boundary knots and the ends of the function are much more constrained than a regular spline or a global polynomoial where the ends can wag a lot more. Natural splines do not require the explicit selection of knots

# There is no need to select the knots here. There is a smoothing parameter which
# can be specified by the degrees of freedom 'df' parameter. The natural spline

fit2=lm(mpg~ns(horsepower,df=4),data=auto)
pred=predict(fit2,newdata=list(horsepower=hpGrid),se=T)
plot(auto$horsepower,auto$mpg,xlim=hp,cex=.5,col="black",xlab="Horsepower",
     ylab="MPG", main="Natural Splines")
lines(hpGrid,pred$fit,lwd=2)
lines(hpGrid,pred$fit+2*pred$se,lty="dashed")
lines(hpGrid,pred$fit-2*pred$se,lty="dashed")

fig3-1

1.1.e Fit a Smoothing Spline – R code

Here a smoothing spline is used. Smoothing splines also do not require the explicit setting of knots. We can change the ‘degrees of freedom(df)’ paramater to get the best fit

# Smoothing spline has a smoothing parameter, the degrees of freedom
# This is too wiggly
plot(auto$horsepower,auto$mpg,xlim=hp,cex=.5,col="black",xlab="Horsepower",
     ylab="MPG", main="Smoothing Splines")

# Here df is set to 16. This has a lot of variance
fit=smooth.spline(auto$horsepower,auto$mpg,df=16)
lines(fit,col="red",lwd=2)

# We can use Cross Validation to allow the spline to pick the value of this smpopothing paramter. We do not need to set the degrees of freedom 'df'
fit=smooth.spline(auto$horsepower,auto$mpg,cv=TRUE)
lines(fit,col="blue",lwd=2)

fig4-1

1.1e Splines – Python

There isn’t as much treatment of splines in Python and SKLearn. I did find the LSQUnivariate, UnivariateSpline spline. The LSQUnivariate spline requires the explcit setting of knots

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from scipy.interpolate import LSQUnivariateSpline
autoDF =pd.read_csv("auto_mpg.csv",encoding="ISO-8859-1")
autoDF.shape
autoDF.columns
autoDF1=autoDF[['mpg','cylinder','displacement','horsepower','weight','acceleration','year']]
autoDF2 = autoDF1.apply(pd.to_numeric, errors='coerce')
auto=autoDF2.dropna()
auto=auto[['horsepower','mpg']].sort_values('horsepower')

# Set the knots manually
knots=[65,75,100,150]
# Create an array for X & y
X=np.array(auto['horsepower'])
y=np.array(auto['mpg'])
# Fit a LSQunivariate spline
s = LSQUnivariateSpline(X,y,knots)

#Plot the spline
xs = np.linspace(40,230,1000)
ys = s(xs)
plt.scatter(X, y)
plt.plot(xs, ys)
plt.savefig('fig2.png', bbox_inches='tight')

fig2

1.2 Generalized Additiive models (GAMs)

Generalized Additive Models (GAMs) is a really powerful ML tool.

y_{i} = \beta_{0} + f_{1}(x_{i1}) + f_{2}(x_{i2}) + .. +f_{p}(x_{ip}) + \epsilon_{i}

In GAMs we use a different functions for each of the variables. GAMs give a much better fit since we can choose any function for the different sections

1.2a Generalized Additive Models (GAMs) – R Code

The plot below show the smooth spline that is fit for each of the features horsepower, cylinder, displacement, year and acceleration. We can use any function for example loess, 4rd order polynomial etc.

library(gam)
# Fit a smoothing spline for horsepower, cyliner, displacement and acceleration
gam=gam(mpg~s(horsepower,4)+s(cylinder,5)+s(displacement,4)+s(year,4)+s(acceleration,5),data=auto)
# Display the summary of the fit. This give the significance of each of the paramwetr
# Also an ANOVA is given for each combination of the features
summary(gam)
## 
## Call: gam(formula = mpg ~ s(horsepower, 4) + s(cylinder, 5) + s(displacement, 
##     4) + s(year, 4) + s(acceleration, 5), data = auto)
## Deviance Residuals:
##     Min      1Q  Median      3Q     Max 
## -8.3190 -1.4436 -0.0261  1.2279 12.0873 
## 
## (Dispersion Parameter for gaussian family taken to be 6.9943)
## 
##     Null Deviance: 23818.99 on 391 degrees of freedom
## Residual Deviance: 2587.881 on 370 degrees of freedom
## AIC: 1898.282 
## 
## Number of Local Scoring Iterations: 3 
## 
## Anova for Parametric Effects
##                     Df  Sum Sq Mean Sq  F value    Pr(>F)    
## s(horsepower, 4)     1 15632.8 15632.8 2235.085 < 2.2e-16 ***
## s(cylinder, 5)       1   508.2   508.2   72.666 3.958e-16 ***
## s(displacement, 4)   1   374.3   374.3   53.514 1.606e-12 ***
## s(year, 4)           1  2263.2  2263.2  323.583 < 2.2e-16 ***
## s(acceleration, 5)   1   372.4   372.4   53.246 1.809e-12 ***
## Residuals          370  2587.9     7.0                       
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Anova for Nonparametric Effects
##                    Npar Df Npar F     Pr(F)    
## (Intercept)                                    
## s(horsepower, 4)         3 13.825 1.453e-08 ***
## s(cylinder, 5)           3 17.668 9.712e-11 ***
## s(displacement, 4)       3 44.573 < 2.2e-16 ***
## s(year, 4)               3 23.364 7.183e-14 ***
## s(acceleration, 5)       4  3.848  0.004453 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
par(mfrow=c(2,3))
plot(gam,se=TRUE)

fig5-1

1.2b Generalized Additive Models (GAMs) – Python Code

I did not find the equivalent of GAMs in SKlearn in Python. There was an early prototype (2012) in Github. Looks like it is still work in progress or has probably been abandoned.

1.3 Tree based Machine Learning Models

Tree based Machine Learning are all based on the ‘bootstrapping’ technique. In bootstrapping given a sample of size N, we create datasets of size N by sampling this original dataset with replacement. Machine Learning models are built on the different bootstrapped samples and then averaged.

Decision Trees as seen above have the tendency to overfit. There are several techniques that help to avoid this namely a) Bagging b) Random Forests c) Boosting

Bagging, Random Forest and Gradient Boosting

Bagging: Bagging, or Bootstrap Aggregation decreases the variance of predictions, by creating separate Decisiion Tree based ML models on the different samples and then averaging these ML models

Random Forests: Bagging is a greedy algorithm and tries to produce splits based on all variables which try to minimize the error. However the different ML models have a high correlation. Random Forests remove this shortcoming, by using a variable and random set of features to split on. Hence the features chosen and the resulting trees are uncorrelated. When these ML models are averaged the performance is much better.

Boosting: Gradient Boosted Decision Trees also use an ensemble of trees but they don’t build Machine Learning models with random set of features at each step. Rather small and simple trees are built. Successive trees try to minimize the error from the earlier trees.

Out of Bag (OOB) Error: In Random Forest and Gradient Boosting for each bootstrap sample taken from the dataset, there will be samples left out. These are known as Out of Bag samples.Classification accuracy carried out on these OOB samples is known as OOB error

1.31a Decision Trees – R Code

The code below creates a Decision tree with the cancer training data. The summary of the fit is output. Based on the ML model, the predict function is used on test data and a confusion matrix is output.

# Read the cancer data
library(tree)
library(caret)
library(e1071)
cancer <- read.csv("cancer.csv",stringsAsFactors = FALSE)
cancer <- cancer[,2:32]
cancer$target <- as.factor(cancer$target)
train_idx <- trainTestSplit(cancer,trainPercent=75,seed=5)
train <- cancer[train_idx, ]
test <- cancer[-train_idx, ]

# Create Decision Tree
cancerStatus=tree(target~.,train)
summary(cancerStatus)
## 
## Classification tree:
## tree(formula = target ~ ., data = train)
## Variables actually used in tree construction:
## [1] "worst.perimeter"      "worst.concave.points" "area.error"          
## [4] "worst.texture"        "mean.texture"         "mean.concave.points" 
## Number of terminal nodes:  9 
## Residual mean deviance:  0.1218 = 50.8 / 417 
## Misclassification error rate: 0.02347 = 10 / 426
pred <- predict(cancerStatus,newdata=test,type="class")
confusionMatrix(pred,test$target)
## Confusion Matrix and Statistics
## 
##           Reference
## Prediction  0  1
##          0 49  7
##          1  8 78
##                                           
##                Accuracy : 0.8944          
##                  95% CI : (0.8318, 0.9397)
##     No Information Rate : 0.5986          
##     P-Value [Acc > NIR] : 4.641e-15       
##                                           
##                   Kappa : 0.7795          
##  Mcnemar's Test P-Value : 1               
##                                           
##             Sensitivity : 0.8596          
##             Specificity : 0.9176          
##          Pos Pred Value : 0.8750          
##          Neg Pred Value : 0.9070          
##              Prevalence : 0.4014          
##          Detection Rate : 0.3451          
##    Detection Prevalence : 0.3944          
##       Balanced Accuracy : 0.8886          
##                                           
##        'Positive' Class : 0               
## 
# Plot decision tree with labels
plot(cancerStatus)
text(cancerStatus,pretty=0)

fig6-1

1.31b Decision Trees – Cross Validation – R Code

We can also perform a Cross Validation on the data to identify the Decision Tree which will give the minimum deviance.

library(tree)
cancer <- read.csv("cancer.csv",stringsAsFactors = FALSE)
cancer <- cancer[,2:32]
cancer$target <- as.factor(cancer$target)
train_idx <- trainTestSplit(cancer,trainPercent=75,seed=5)
train <- cancer[train_idx, ]
test <- cancer[-train_idx, ]

# Create Decision Tree
cancerStatus=tree(target~.,train)

# Execute 10 fold cross validation
cvCancer=cv.tree(cancerStatus)
plot(cvCancer)

fig7-1

# Plot the 
plot(cvCancer$size,cvCancer$dev,type='b')

fig1

prunedCancer=prune.tree(cancerStatus,best=4)
plot(prunedCancer)
text(prunedCancer,pretty=0)

fig2

pred <- predict(prunedCancer,newdata=test,type="class")
confusionMatrix(pred,test$target)
## Confusion Matrix and Statistics
## 
##           Reference
## Prediction  0  1
##          0 50  7
##          1  7 78
##                                          
##                Accuracy : 0.9014         
##                  95% CI : (0.8401, 0.945)
##     No Information Rate : 0.5986         
##     P-Value [Acc > NIR] : 7.988e-16      
##                                          
##                   Kappa : 0.7948         
##  Mcnemar's Test P-Value : 1              
##                                          
##             Sensitivity : 0.8772         
##             Specificity : 0.9176         
##          Pos Pred Value : 0.8772         
##          Neg Pred Value : 0.9176         
##              Prevalence : 0.4014         
##          Detection Rate : 0.3521         
##    Detection Prevalence : 0.4014         
##       Balanced Accuracy : 0.8974         
##                                          
##        'Positive' Class : 0              
## 

1.31c Decision Trees – Python Code

Below is the Python code for creating Decision Trees. The accuracy, precision, recall and F1 score is computed on the test data set.

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix
from sklearn import tree
from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.datasets import make_classification, make_blobs
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score
import graphviz 

cancer = load_breast_cancer()
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)

X_train, X_test, y_train, y_test = train_test_split(X_cancer, y_cancer,
                                                   random_state = 0)
clf = DecisionTreeClassifier().fit(X_train, y_train)

print('Accuracy of Decision Tree classifier on training set: {:.2f}'
     .format(clf.score(X_train, y_train)))
print('Accuracy of Decision Tree classifier on test set: {:.2f}'
     .format(clf.score(X_test, y_test)))

y_predicted=clf.predict(X_test)
confusion = confusion_matrix(y_test, y_predicted)
print('Accuracy: {:.2f}'.format(accuracy_score(y_test, y_predicted)))
print('Precision: {:.2f}'.format(precision_score(y_test, y_predicted)))
print('Recall: {:.2f}'.format(recall_score(y_test, y_predicted)))
print('F1: {:.2f}'.format(f1_score(y_test, y_predicted)))

# Plot the Decision Tree
clf = DecisionTreeClassifier(max_depth=2).fit(X_train, y_train)
dot_data = tree.export_graphviz(clf, out_file=None, 
                         feature_names=cancer.feature_names,  
                         class_names=cancer.target_names,  
                         filled=True, rounded=True,  
                         special_characters=True)  
graph = graphviz.Source(dot_data)  
graph
## Accuracy of Decision Tree classifier on training set: 1.00
## Accuracy of Decision Tree classifier on test set: 0.87
## Accuracy: 0.87
## Precision: 0.97
## Recall: 0.82
## F1: 0.89

tree

1.31d Decision Trees – Cross Validation – Python Code

In the code below 5-fold cross validation is performed for different depths of the tree and the accuracy is computed. The accuracy on the test set seems to plateau when the depth is 8. But it is seen to increase again from 10 to 12. More analysis needs to be done here


import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.datasets import load_breast_cancer
from sklearn.tree import DecisionTreeClassifier
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)
from sklearn.cross_validation import train_test_split, KFold
def computeCVAccuracy(X,y,folds):
    accuracy=[]
    foldAcc=[]
    depth=[1,2,3,4,5,6,7,8,9,10,11,12]
    nK=len(X)/float(folds)
    xval_err=0
    for i in depth: 
        kf = KFold(len(X),n_folds=folds)
        for train_index, test_index in kf:
            X_train, X_test = X.iloc[train_index], X.iloc[test_index]
            y_train, y_test = y.iloc[train_index], y.iloc[test_index]  
            clf = DecisionTreeClassifier(max_depth = i).fit(X_train, y_train)
            score=clf.score(X_test, y_test)
            accuracy.append(score)     
            
        foldAcc.append(np.mean(accuracy))  
        
    return(foldAcc)
    
    
cvAccuracy=computeCVAccuracy(pd.DataFrame(X_cancer),pd.DataFrame(y_cancer),folds=10)

df1=pd.DataFrame(cvAccuracy)
df1.columns=['cvAccuracy']
df=df1.reindex([1,2,3,4,5,6,7,8,9,10,11,12])
df.plot()
plt.title("Decision Tree - 10-fold Cross Validation Accuracy vs Depth of tree")
plt.xlabel("Depth of tree")
plt.ylabel("Accuracy")
plt.savefig('fig3.png', bbox_inches='tight')

 

 

fig3

 

1.4a Random Forest – R code

A Random Forest is fit using the Boston data. The summary shows that 4 variables were randomly chosen at each split and the resulting ML model explains 88.72% of the test data. Also the variable importance is plotted. It can be seen that ‘rooms’ and ‘status’ are the most influential features in the model

library(randomForest)
df=read.csv("Boston.csv",stringsAsFactors = FALSE) # Data from MASS - SL

# Select specific columns
Boston <- df %>% dplyr::select("crimeRate","zone","indus","charles","nox","rooms","age",                          "distances","highways","tax","teacherRatio","color",
                               "status","medianValue")

# Fit a Random Forest on the Boston training data
rfBoston=randomForest(medianValue~.,data=Boston)
# Display the summatu of the fit. It can be seen that the MSE is 10.88 
# and the percentage variance explained is 86.14%. About 4 variables were tried at each # #split for a maximum tree of 500.
# The MSE and percent variance is on Out of Bag trees
rfBoston
## 
## Call:
##  randomForest(formula = medianValue ~ ., data = Boston) 
##                Type of random forest: regression
##                      Number of trees: 500
## No. of variables tried at each split: 4
## 
##           Mean of squared residuals: 9.521672
##                     % Var explained: 88.72
#List and plot the variable importances
importance(rfBoston)
##              IncNodePurity
## crimeRate        2602.1550
## zone              258.8057
## indus            2599.6635
## charles           240.2879
## nox              2748.8485
## rooms           12011.6178
## age              1083.3242
## distances        2432.8962
## highways          393.5599
## tax              1348.6987
## teacherRatio     2841.5151
## color             731.4387
## status          12735.4046
varImpPlot(rfBoston)

fig8-1

1.4b Random Forest-OOB and Cross Validation Error – R code

The figure below shows the OOB error and the Cross Validation error vs the ‘mtry’. Here mtry indicates the number of random features that are chosen at each split. The lowest test error occurs when mtry = 8

library(randomForest)
df=read.csv("Boston.csv",stringsAsFactors = FALSE) # Data from MASS - SL

# Select specific columns
Boston <- df %>% dplyr::select("crimeRate","zone","indus","charles","nox","rooms","age",                          "distances","highways","tax","teacherRatio","color",
                               "status","medianValue")
# Split as training and tst sets
train_idx <- trainTestSplit(Boston,trainPercent=75,seed=5)
train <- Boston[train_idx, ]
test <- Boston[-train_idx, ]

#Initialize OOD and testError
oobError <- NULL
testError <- NULL
# In the code below the number of variables to consider at each split is increased
# from 1 - 13(max features) and the OOB error and the MSE is computed
for(i in 1:13){
    fitRF=randomForest(medianValue~.,data=train,mtry=i,ntree=400)
    oobError[i] <-fitRF$mse[400]
    pred <- predict(fitRF,newdata=test)
    testError[i] <- mean((pred-test$medianValue)^2)
}

# We can see the OOB and Test Error. It can be seen that the Random Forest performs
# best with the lowers MSE at mtry=6
matplot(1:13,cbind(testError,oobError),pch=19,col=c("red","blue"),
        type="b",xlab="mtry(no of varaibles at each split)", ylab="Mean Squared Error",
        main="Random Forest - OOB and Test Error")
legend("topright",legend=c("OOB","Test"),pch=19,col=c("red","blue"))

fig9-1

1.4c Random Forest – Python code

The python code for Random Forest Regression is shown below. The training and test score is computed. The variable importance shows that ‘rooms’ and ‘status’ are the most influential of the variables

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestRegressor
df = pd.read_csv("Boston.csv",encoding = "ISO-8859-1")

X=df[['crimeRate','zone', 'indus','charles','nox','rooms', 'age','distances','highways','tax',
       'teacherRatio','color','status']]
y=df['medianValue']

X_train, X_test, y_train, y_test = train_test_split(X, y, random_state = 0)

regr = RandomForestRegressor(max_depth=4, random_state=0)
regr.fit(X_train, y_train)

print('R-squared score (training): {:.3f}'
     .format(regr.score(X_train, y_train)))
print('R-squared score (test): {:.3f}'
     .format(regr.score(X_test, y_test)))

feature_names=['crimeRate','zone', 'indus','charles','nox','rooms', 'age','distances','highways','tax',
       'teacherRatio','color','status']
print(regr.feature_importances_)
plt.figure(figsize=(10,6),dpi=80)
c_features=X_train.shape[1]
plt.barh(np.arange(c_features),regr.feature_importances_)
plt.xlabel("Feature importance")
plt.ylabel("Feature name")

plt.yticks(np.arange(c_features), feature_names)
plt.tight_layout()

plt.savefig('fig4.png', bbox_inches='tight')
## R-squared score (training): 0.917
## R-squared score (test): 0.734
## [ 0.03437382  0.          0.00580335  0.          0.00731004  0.36461548
##   0.00638577  0.03432173  0.0041244   0.01732328  0.01074148  0.0012638
##   0.51373683]

fig4

1.4d Random Forest – Cross Validation and OOB Error – Python code

As with R the ‘max_features’ determines the random number of features the random forest will use at each split. The plot shows that when max_features=8 the MSE is lowest

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import cross_val_score
df = pd.read_csv("Boston.csv",encoding = "ISO-8859-1")

X=df[['crimeRate','zone', 'indus','charles','nox','rooms', 'age','distances','highways','tax',
       'teacherRatio','color','status']]
y=df['medianValue']

cvError=[]
oobError=[]
oobMSE=[]
for i in range(1,13):
    regr = RandomForestRegressor(max_depth=4, n_estimators=400,max_features=i,oob_score=True,random_state=0)
    mse= np.mean(cross_val_score(regr, X, y, cv=5,scoring = 'neg_mean_squared_error'))
    # Since this is neg_mean_squared_error I have inverted the sign to get MSE
    cvError.append(-mse)
    # Fit on all data to compute OOB error
    regr.fit(X, y)
    # Record the OOB error for each `max_features=i` setting
    oob = 1 - regr.oob_score_
    oobError.append(oob)
    # Get the Out of Bag prediction
    oobPred=regr.oob_prediction_ 
    # Compute the Mean Squared Error between OOB Prediction and target
    mseOOB=np.mean(np.square(oobPred-y))
    oobMSE.append(mseOOB)

# Plot the CV Error and OOB Error
# Set max_features
maxFeatures=np.arange(1,13) 
cvError=pd.DataFrame(cvError,index=maxFeatures)
oobMSE=pd.DataFrame(oobMSE,index=maxFeatures)
#Plot
fig8=df.plot()
fig8=plt.title('Random forest - CV Error and OOB Error vs max_features')
fig8.figure.savefig('fig8.png', bbox_inches='tight')

#Plot the OOB Error vs max_features
plt.plot(range(1,13),oobError)
fig2=plt.title("Random Forest - OOB Error vs max_features (variable no of features)")
fig2=plt.xlabel("max_features (variable no of features)")
fig2=plt.ylabel("OOB Error")
fig2.figure.savefig('fig7.png', bbox_inches='tight')

fig8 fig7

1.5a Boosting – R code

Here a Gradient Boosted ML Model is built with a n.trees=5000, with a learning rate of 0.01 and depth of 4. The feature importance plot also shows that rooms and status are the 2 most important features. The MSE vs the number of trees plateaus around 2000 trees

library(gbm)
# Perform gradient boosting on the Boston data set. The distribution is gaussian since we
# doing MSE. The interaction depth specifies the number of splits
boostBoston=gbm(medianValue~.,data=train,distribution="gaussian",n.trees=5000,
                shrinkage=0.01,interaction.depth=4)
#The summary gives the variable importance. The 2 most significant variables are
# number of rooms and lower status
summary(boostBoston)

##                       var    rel.inf
## rooms               rooms 42.2267200
## status             status 27.3024671
## distances       distances  7.9447972
## crimeRate       crimeRate  5.0238827
## nox                   nox  4.0616548
## teacherRatio teacherRatio  3.1991999
## age                   age  2.7909772
## color               color  2.3436295
## tax                   tax  2.1386213
## charles           charles  1.3799109
## highways         highways  0.7644026
## indus               indus  0.7236082
## zone                 zone  0.1001287
# The plots below show how each variable relates to the median value of the home. As
# the number of roomd increase the median value increases and with increase in lower status
# the median value decreases
par(mfrow=c(1,2))
#Plot the relation between the top 2 features and the target
plot(boostBoston,i="rooms")
plot(boostBoston,i="status")

fig10-2

# Create a sequence of trees between 100-5000 incremented by 50
nTrees=seq(100,5000,by=50)
# Predict the values for the test data
pred <- predict(boostBoston,newdata=test,n.trees=nTrees)
# Compute the mean for each of the MSE for each of the number of trees 
boostError <- apply((pred-test$medianValue)^2,2,mean)
#Plot the MSE vs the number of trees
plot(nTrees,boostError,pch=19,col="blue",ylab="Mean Squared Error",
     main="Boosting Test Error")

fig10-3

1.5b Cross Validation Boosting – R code

Included below is a cross validation error vs the learning rate. The lowest error is when learning rate = 0.09

cvError <- NULL
s <- c(.001,0.01,0.03,0.05,0.07,0.09,0.1)
for(i in seq_along(s)){
    cvBoost=gbm(medianValue~.,data=train,distribution="gaussian",n.trees=5000,
                shrinkage=s[i],interaction.depth=4,cv.folds=5)
    cvError[i] <- mean(cvBoost$cv.error)
}

# Create a data frame for plotting
a <- rbind(s,cvError)
b <- as.data.frame(t(a))
# It can be seen that a shrinkage parameter of 0,05 gives the lowes CV Error
ggplot(b,aes(s,cvError)) + geom_point() + geom_line(color="blue") + 
    xlab("Shrinkage") + ylab("Cross Validation Error") +
    ggtitle("Gradient boosted trees - Cross Validation error vs Shrinkage")

fig11-1

1.5c Boosting – Python code

A gradient boost ML model in Python is created below. The Rsquared score is computed on the training and test data.

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.ensemble import GradientBoostingRegressor
df = pd.read_csv("Boston.csv",encoding = "ISO-8859-1")

X=df[['crimeRate','zone', 'indus','charles','nox','rooms', 'age','distances','highways','tax',
       'teacherRatio','color','status']]
y=df['medianValue']

X_train, X_test, y_train, y_test = train_test_split(X, y, random_state = 0)

regr = GradientBoostingRegressor()
regr.fit(X_train, y_train)

print('R-squared score (training): {:.3f}'
     .format(regr.score(X_train, y_train)))
print('R-squared score (test): {:.3f}'
     .format(regr.score(X_test, y_test)))
## R-squared score (training): 0.983
## R-squared score (test): 0.821

1.5c Cross Validation Boosting – Python code

the cross validation error is computed as the learning rate is varied. The minimum CV eror occurs when lr = 0.04

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestRegressor
from sklearn.ensemble import GradientBoostingRegressor
from sklearn.model_selection import cross_val_score
df = pd.read_csv("Boston.csv",encoding = "ISO-8859-1")

X=df[['crimeRate','zone', 'indus','charles','nox','rooms', 'age','distances','highways','tax',
       'teacherRatio','color','status']]
y=df['medianValue']

cvError=[]
learning_rate =[.001,0.01,0.03,0.05,0.07,0.09,0.1]
for lr in learning_rate:
    regr = GradientBoostingRegressor(max_depth=4, n_estimators=400,learning_rate  =lr,random_state=0)
    mse= np.mean(cross_val_score(regr, X, y, cv=10,scoring = 'neg_mean_squared_error'))
    # Since this is neg_mean_squared_error I have inverted the sign to get MSE
    cvError.append(-mse)
learning_rate =[.001,0.01,0.03,0.05,0.07,0.09,0.1]
plt.plot(learning_rate,cvError)
plt.title("Gradient Boosting - 5-fold CV- Mean Squared Error vs max_features (variable no of features)")
plt.xlabel("max_features (variable no of features)")
plt.ylabel("Mean Squared Error")
plt.savefig('fig6.png', bbox_inches='tight')

fig6

Conclusion This post covered Splines and Tree based ML models like Bagging, Random Forest and Boosting. Stay tuned for further updates.

You may also like

  1. Re-introducing cricketr! : An R package to analyze performances of cricketer
  2. Designing a Social Web Portal
  3. My travels through the realms of Data Science, Machine Learning, Deep Learning and (AI)
  4. My TEDx talk on the “Internet of Things”
  5. Singularity
  6. A closer look at “Robot Horse on a Trot” in Android

To see all posts see Index of posts

Advertisements

Practical Machine Learning with R and Python – Part 4


This is the 4th installment of my ‘Practical Machine Learning with R and Python’ series. In this part I discuss classification with Support Vector Machines (SVMs), using both a Linear and a Radial basis kernel, and Decision Trees. Further, a closer look is taken at some of the metrics associated with binary classification, namely accuracy vs precision and recall. I also touch upon Validation curves, Precision-Recall, ROC curves and AUC with equivalent code in R and Python

This post is a continuation of my 3 earlier posts on Practical Machine Learning in R and Python
1. Practical Machine Learning with R and Python – Part 1
2. Practical Machine Learning with R and Python – Part 2
3. Practical Machine Learning with R and Python – Part 3

The RMarkdown file with the code and the associated data files can be downloaded from Github at MachineLearning-RandPython-Part4

The content of this post and much more is now available as a compact book  on Amazon in both formats – as Paperback ($9.99) and a Kindle version($6.99/Rs449/). see ‘Practical Machine Learning with R and Python – Machine Learning in stereo

Support Vector Machines (SVM) are another useful Machine Learning model that can be used for both regression and classification problems. SVMs used in classification, compute the hyperplane, that separates the 2 classes with the maximum margin. To do this the features may be transformed into a larger multi-dimensional feature space. SVMs can be used with different kernels namely linear, polynomial or radial basis to determine the best fitting model for a given classification problem.

In the 2nd part of this series Practical Machine Learning with R and Python – Part 2, I had mentioned the various metrics that are used in classification ML problems namely Accuracy, Precision, Recall and F1 score. Accuracy gives the fraction of data that were correctly classified as belonging to the +ve or -ve class. However ‘accuracy’ in itself is not a good enough measure because it does not take into account the fraction of the data that were incorrectly classified. This issue becomes even more critical in different domains. For e.g a surgeon who would like to detect cancer, would like to err on the side of caution, and classify even a possibly non-cancerous patient as possibly having cancer, rather than mis-classifying a malignancy as benign. Here we would like to increase recall or sensitivity which is  given by Recall= TP/(TP+FN) or we try reduce mis-classification by either increasing the (true positives) TP or reducing (false negatives) FN

On the other hand, search algorithms would like to increase precision which tries to reduce the number of irrelevant results in the search result. Precision= TP/(TP+FP). In other words we do not want ‘false positives’ or irrelevant results to come in the search results and there is a need to reduce the false positives.

When we try to increase ‘precision’, we do so at the cost of ‘recall’, and vice-versa. I found this diagram and explanation in Wikipedia very useful Source: Wikipedia

“Consider a brain surgeon tasked with removing a cancerous tumor from a patient’s brain. The surgeon needs to remove all of the tumor cells since any remaining cancer cells will regenerate the tumor. Conversely, the surgeon must not remove healthy brain cells since that would leave the patient with impaired brain function. The surgeon may be more liberal in the area of the brain she removes to ensure she has extracted all the cancer cells. This decision increases recall but reduces precision. On the other hand, the surgeon may be more conservative in the brain she removes to ensure she extracts only cancer cells. This decision increases precision but reduces recall. That is to say, greater recall increases the chances of removing healthy cells (negative outcome) and increases the chances of removing all cancer cells (positive outcome). Greater precision decreases the chances of removing healthy cells (positive outcome) but also decreases the chances of removing all cancer cells (negative outcome).”

1.1a. Linear SVM – R code

In R code below I use SVM with linear kernel

source('RFunctions-1.R')
library(dplyr)
library(e1071)
library(caret)
library(reshape2)
library(ggplot2)
# Read data. Data from SKLearn
cancer <- read.csv("cancer.csv")
cancer$target <- as.factor(cancer$target)

# Split into training and test sets
train_idx <- trainTestSplit(cancer,trainPercent=75,seed=5)
train <- cancer[train_idx, ]
test <- cancer[-train_idx, ]

# Fit a linear basis kernel. DO not scale the data
svmfit=svm(target~., data=train, kernel="linear",scale=FALSE)
ypred=predict(svmfit,test)
#Print a confusion matrix
confusionMatrix(ypred,test$target)
## Confusion Matrix and Statistics
## 
##           Reference
## Prediction  0  1
##          0 54  3
##          1  3 82
##                                           
##                Accuracy : 0.9577          
##                  95% CI : (0.9103, 0.9843)
##     No Information Rate : 0.5986          
##     P-Value [Acc > NIR] : <2e-16          
##                                           
##                   Kappa : 0.9121          
##  Mcnemar's Test P-Value : 1               
##                                           
##             Sensitivity : 0.9474          
##             Specificity : 0.9647          
##          Pos Pred Value : 0.9474          
##          Neg Pred Value : 0.9647          
##              Prevalence : 0.4014          
##          Detection Rate : 0.3803          
##    Detection Prevalence : 0.4014          
##       Balanced Accuracy : 0.9560          
##                                           
##        'Positive' Class : 0               
## 

1.1b Linear SVM – Python code

The code below creates a SVM with linear basis in Python and also dumps the corresponding classification metrics

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.svm import LinearSVC

from sklearn.datasets import make_classification, make_blobs

from sklearn.metrics import confusion_matrix
from matplotlib.colors import ListedColormap
from sklearn.datasets import load_breast_cancer
# Load the cancer data
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)
X_train, X_test, y_train, y_test = train_test_split(X_cancer, y_cancer,
                                                   random_state = 0)
clf = LinearSVC().fit(X_train, y_train)
print('Breast cancer dataset')
print('Accuracy of Linear SVC classifier on training set: {:.2f}'
     .format(clf.score(X_train, y_train)))
print('Accuracy of Linear SVC classifier on test set: {:.2f}'
     .format(clf.score(X_test, y_test)))
## Breast cancer dataset
## Accuracy of Linear SVC classifier on training set: 0.92
## Accuracy of Linear SVC classifier on test set: 0.94

1.2 Dummy classifier

Often when we perform classification tasks using any ML model namely logistic regression, SVM, neural networks etc. it is very useful to determine how well the ML model performs agains at dummy classifier. A dummy classifier uses some simple computation like frequency of majority class, instead of fitting and ML model. It is essential that our ML model does much better that the dummy classifier. This problem is even more important in imbalanced classes where we have only about 10% of +ve samples. If any ML model we create has a accuracy of about 0.90 then it is evident that our classifier is not doing any better than a dummy classsfier which can just take a majority count of this imbalanced class and also come up with 0.90. We need to be able to do better than that.

In the examples below (1.3a & 1.3b) it can be seen that SVMs with ‘radial basis’ kernel with unnormalized data, for both R and Python, do not perform any better than the dummy classifier.

1.2a Dummy classifier – R code

R does not seem to have an explicit dummy classifier. I created a simple dummy classifier that predicts the majority class. SKlearn in Python also includes other strategies like uniform, stratified etc. but this should be possible to create in R also.

# Create a simple dummy classifier that computes the ratio of the majority class to the totla
DummyClassifierAccuracy <- function(train,test,type="majority"){
  if(type=="majority"){
      count <- sum(train$target==1)/dim(train)[1]
  }
  count
}


cancer <- read.csv("cancer.csv")
cancer$target <- as.factor(cancer$target)

# Create training and test sets
train_idx <- trainTestSplit(cancer,trainPercent=75,seed=5)
train <- cancer[train_idx, ]
test <- cancer[-train_idx, ]

#Dummy classifier majority class
acc=DummyClassifierAccuracy(train,test)
sprintf("Accuracy is %f",acc)
## [1] "Accuracy is 0.638498"

1.2b Dummy classifier – Python code

This dummy classifier uses the majority class.

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.dummy import DummyClassifier
from sklearn.metrics import confusion_matrix
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)
X_train, X_test, y_train, y_test = train_test_split(X_cancer, y_cancer,
                                                   random_state = 0)

# Negative class (0) is most frequent
dummy_majority = DummyClassifier(strategy = 'most_frequent').fit(X_train, y_train)
y_dummy_predictions = dummy_majority.predict(X_test)

print('Dummy classifier accuracy on test set: {:.2f}'
     .format(dummy_majority.score(X_test, y_test)))
## Dummy classifier accuracy on test set: 0.63

1.3a – Radial SVM (un-normalized) – R code

SVMs perform better when the data is normalized or scaled. The 2 examples below show that SVM with radial basis kernel does not perform any better than the dummy classifier

library(dplyr)
library(e1071)
library(caret)
library(reshape2)
library(ggplot2)

# Radial SVM unnormalized
train_idx <- trainTestSplit(cancer,trainPercent=75,seed=5)
train <- cancer[train_idx, ]
test <- cancer[-train_idx, ]
# Unnormalized data
svmfit=svm(target~., data=train, kernel="radial",cost=10,scale=FALSE)
ypred=predict(svmfit,test)
confusionMatrix(ypred,test$target)
## Confusion Matrix and Statistics
## 
##           Reference
## Prediction  0  1
##          0  0  0
##          1 57 85
##                                           
##                Accuracy : 0.5986          
##                  95% CI : (0.5131, 0.6799)
##     No Information Rate : 0.5986          
##     P-Value [Acc > NIR] : 0.5363          
##                                           
##                   Kappa : 0               
##  Mcnemar's Test P-Value : 1.195e-13       
##                                           
##             Sensitivity : 0.0000          
##             Specificity : 1.0000          
##          Pos Pred Value :    NaN          
##          Neg Pred Value : 0.5986          
##              Prevalence : 0.4014          
##          Detection Rate : 0.0000          
##    Detection Prevalence : 0.0000          
##       Balanced Accuracy : 0.5000          
##                                           
##        'Positive' Class : 0               
## 

1.4b – Radial SVM (un-normalized) – Python code

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.svm import SVC

# Load the cancer data
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)
X_train, X_test, y_train, y_test = train_test_split(X_cancer, y_cancer,
                                                   random_state = 0)


clf = SVC(C=10).fit(X_train, y_train)
print('Breast cancer dataset (unnormalized features)')
print('Accuracy of RBF-kernel SVC on training set: {:.2f}'
     .format(clf.score(X_train, y_train)))
print('Accuracy of RBF-kernel SVC on test set: {:.2f}'
     .format(clf.score(X_test, y_test)))
## Breast cancer dataset (unnormalized features)
## Accuracy of RBF-kernel SVC on training set: 1.00
## Accuracy of RBF-kernel SVC on test set: 0.63

1.5a – Radial SVM (Normalized) -R Code

The data is scaled (normalized ) before using the SVM model. The SVM model has 2 paramaters a) C – Large C (less regularization), more regularization b) gamma – Small gamma has larger decision boundary with more misclassfication, and larger gamma has tighter decision boundary

The R code below computes the accuracy as the regularization paramater is changed

trainingAccuracy <- NULL
testAccuracy <- NULL
C1 <- c(.01,.1, 1, 10, 20)
for(i in  C1){
  
    svmfit=svm(target~., data=train, kernel="radial",cost=i,scale=TRUE)
    ypredTrain <-predict(svmfit,train)
    ypredTest=predict(svmfit,test)
    a <-confusionMatrix(ypredTrain,train$target)
    b <-confusionMatrix(ypredTest,test$target)
    trainingAccuracy <-c(trainingAccuracy,a$overall[1])
    testAccuracy <-c(testAccuracy,b$overall[1])
    
}
print(trainingAccuracy)
##  Accuracy  Accuracy  Accuracy  Accuracy  Accuracy 
## 0.6384977 0.9671362 0.9906103 0.9976526 1.0000000
print(testAccuracy)
##  Accuracy  Accuracy  Accuracy  Accuracy  Accuracy 
## 0.5985915 0.9507042 0.9647887 0.9507042 0.9507042
a <-rbind(C1,as.numeric(trainingAccuracy),as.numeric(testAccuracy))
b <- data.frame(t(a))
names(b) <- c("C1","trainingAccuracy","testAccuracy")
df <- melt(b,id="C1")
ggplot(df) + geom_line(aes(x=C1, y=value, colour=variable),size=2) +
    xlab("C (SVC regularization)value") + ylab("Accuracy") +
    ggtitle("Training and test accuracy vs C(regularization)")

1.5b – Radial SVM (normalized) – Python

The Radial basis kernel is used on normalized data for a range of ‘C’ values and the result is plotted.

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.svm import SVC
from sklearn.preprocessing import MinMaxScaler
scaler = MinMaxScaler()

# Load the cancer data
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)
X_train, X_test, y_train, y_test = train_test_split(X_cancer, y_cancer,
                                                   random_state = 0)
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)
   
print('Breast cancer dataset (normalized with MinMax scaling)')
trainingAccuracy=[]
testAccuracy=[]
for C1 in [.01,.1, 1, 10, 20]:
    clf = SVC(C=C1).fit(X_train_scaled, y_train)
    acctrain=clf.score(X_train_scaled, y_train)
    accTest=clf.score(X_test_scaled, y_test)
    trainingAccuracy.append(acctrain)
    testAccuracy.append(accTest)
    
# Create a dataframe
C1=[.01,.1, 1, 10, 20]   
trainingAccuracy=pd.DataFrame(trainingAccuracy,index=C1)
testAccuracy=pd.DataFrame(testAccuracy,index=C1)

# Plot training and test R squared as a function of alpha
df=pd.concat([trainingAccuracy,testAccuracy],axis=1)
df.columns=['trainingAccuracy','trainingAccuracy']

fig1=df.plot()
fig1=plt.title('Training and test accuracy vs C (SVC)')
fig1.figure.savefig('fig1.png', bbox_inches='tight')
## Breast cancer dataset (normalized with MinMax scaling)

Output image: 

1.6a Validation curve – R code

Sklearn includes code creating validation curves by varying paramaters and computing and plotting accuracy as gamma or C or changd. I did not find this R but I think this is a useful function and so I have created the R equivalent of this.

# The R equivalent of np.logspace
seqLogSpace <- function(start,stop,len){
  a=seq(log10(10^start),log10(10^stop),length=len)
  10^a
}

# Read the data. This is taken the SKlearn cancer data
cancer <- read.csv("cancer.csv")
cancer$target <- as.factor(cancer$target)

set.seed(6)

# Create the range of C1 in log space
param_range = seqLogSpace(-3,2,20)
# Initialize the overall training and test accuracy to NULL
overallTrainAccuracy <- NULL
overallTestAccuracy <- NULL

# Loop over the parameter range of Gamma
for(i in param_range){
    # Set no of folds
    noFolds=5
    # Create the rows which fall into different folds from 1..noFolds
    folds = sample(1:noFolds, nrow(cancer), replace=TRUE) 
    # Initialize the training and test accuracy of folds to 0
    trainingAccuracy <- 0
    testAccuracy <- 0
    
    # Loop through the folds
    for(j in 1:noFolds){
        # The training is all rows for which the row is != j (k-1 folds -> training)
        train <- cancer[folds!=j,]
        # The rows which have j as the index become the test set
        test <- cancer[folds==j,]
        # Create a SVM model for this
        svmfit=svm(target~., data=train, kernel="radial",gamma=i,scale=TRUE)
  
        # Add up all the fold accuracy for training and test separately  
        ypredTrain <-predict(svmfit,train)
        ypredTest=predict(svmfit,test)
        
        # Create confusion matrix 
        a <-confusionMatrix(ypredTrain,train$target)
        b <-confusionMatrix(ypredTest,test$target)
        # Get the accuracy
        trainingAccuracy <-trainingAccuracy + a$overall[1]
        testAccuracy <-testAccuracy+b$overall[1]

    }
    # Compute the average of accuracy for K folds for number of features 'i'
    overallTrainAccuracy=c(overallTrainAccuracy,trainingAccuracy/noFolds)
    overallTestAccuracy=c(overallTestAccuracy,testAccuracy/noFolds)
}
#Create a dataframe
a <- rbind(param_range,as.numeric(overallTrainAccuracy),
               as.numeric(overallTestAccuracy))
b <- data.frame(t(a))
names(b) <- c("C1","trainingAccuracy","testAccuracy")
df <- melt(b,id="C1")
#Plot in log axis
ggplot(df) + geom_line(aes(x=C1, y=value, colour=variable),size=2) +
      xlab("C (SVC regularization)value") + ylab("Accuracy") +
      ggtitle("Training and test accuracy vs C(regularization)") + scale_x_log10()

1.6b Validation curve – Python

Compute and plot the validation curve as gamma is varied.

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import MinMaxScaler
from sklearn.svm import SVC
from sklearn.model_selection import validation_curve


# Load the cancer data
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)
scaler = MinMaxScaler()
X_scaled = scaler.fit_transform(X_cancer)

# Create a gamma values from 10^-3 to 10^2 with 20 equally spaced intervals
param_range = np.logspace(-3, 2, 20)
# Compute the validation curve
train_scores, test_scores = validation_curve(SVC(), X_scaled, y_cancer,
                                            param_name='gamma',
                                            param_range=param_range, cv=10)
                                            
#Plot the figure                                           
fig2=plt.figure()

#Compute the mean
train_scores_mean = np.mean(train_scores, axis=1)
train_scores_std = np.std(train_scores, axis=1)
test_scores_mean = np.mean(test_scores, axis=1)
test_scores_std = np.std(test_scores, axis=1)

fig2=plt.title('Validation Curve with SVM')
fig2=plt.xlabel('$\gamma$ (gamma)')
fig2=plt.ylabel('Score')
fig2=plt.ylim(0.0, 1.1)
lw = 2

fig2=plt.semilogx(param_range, train_scores_mean, label='Training score',
            color='darkorange', lw=lw)

fig2=plt.fill_between(param_range, train_scores_mean - train_scores_std,
                train_scores_mean + train_scores_std, alpha=0.2,
                color='darkorange', lw=lw)

fig2=plt.semilogx(param_range, test_scores_mean, label='Cross-validation score',
            color='navy', lw=lw)

fig2=plt.fill_between(param_range, test_scores_mean - test_scores_std,
                test_scores_mean + test_scores_std, alpha=0.2,
                color='navy', lw=lw)
fig2.figure.savefig('fig2.png', bbox_inches='tight')

Output image: 

1.7a Validation Curve (Preventing data leakage) – Python code

In this course Applied Machine Learning in Python, the Professor states that when we apply the same data transformation to a entire dataset, it will cause a data leakage. “The proper way to do cross-validation when you need to scale the data is not to scale the entire dataset with a single transform, since this will indirectly leak information into the training data about the whole dataset, including the test data (see the lecture on data leakage later in the course). Instead, scaling/normalizing must be computed and applied for each cross-validation fold separately”

So I apply separate scaling to the training and testing folds and plot. In the lecture the Prof states that this can be done using pipelines.

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.datasets import load_breast_cancer
from sklearn.cross_validation import  KFold
from sklearn.preprocessing import MinMaxScaler
from sklearn.svm import SVC

# Read the data
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)
# Set the parameter range
param_range = np.logspace(-3, 2, 20)

# Set number of folds
folds=5
#Initialize
overallTrainAccuracy=[]
overallTestAccuracy=[]

# Loop over the paramater range
for c in  param_range:
    trainingAccuracy=0
    testAccuracy=0
    kf = KFold(len(X_cancer),n_folds=folds)
    # Partition into training and test folds
    for train_index, test_index in kf:
            # Partition the data acccording the fold indices generated
            X_train, X_test = X_cancer[train_index], X_cancer[test_index]
            y_train, y_test = y_cancer[train_index], y_cancer[test_index]  

            
            # Scale the X_train and X_test 
            scaler = MinMaxScaler()
            X_train_scaled = scaler.fit_transform(X_train)
            X_test_scaled = scaler.transform(X_test)
            # Fit a SVC model for each C
            clf = SVC(C=c).fit(X_train_scaled, y_train)
            #Compute the training and test score
            acctrain=clf.score(X_train_scaled, y_train)
            accTest=clf.score(X_test_scaled, y_test)
            trainingAccuracy += np.sum(acctrain)
            testAccuracy += np.sum(accTest)
    # Compute the mean training and testing accuracy
    overallTrainAccuracy.append(trainingAccuracy/folds)
    overallTestAccuracy.append(testAccuracy/folds)
        

overallTrainAccuracy=pd.DataFrame(overallTrainAccuracy,index=param_range)
overallTestAccuracy=pd.DataFrame(overallTestAccuracy,index=param_range)

# Plot training and test R squared as a function of alpha
df=pd.concat([overallTrainAccuracy,overallTestAccuracy],axis=1)
df.columns=['trainingAccuracy','testAccuracy']


fig3=plt.title('Validation Curve with SVM')
fig3=plt.xlabel('$\gamma$ (gamma)')
fig3=plt.ylabel('Score')
fig3=plt.ylim(0.5, 1.1)
lw = 2

fig3=plt.semilogx(param_range, overallTrainAccuracy, label='Training score',
            color='darkorange', lw=lw)

fig3=plt.semilogx(param_range, overallTestAccuracy, label='Cross-validation score',
            color='navy', lw=lw)

fig3=plt.legend(loc='best')
fig3.figure.savefig('fig3.png', bbox_inches='tight')

Output image: 

1.8 a Decision trees – R code

Decision trees in R can be plotted using RPart package

library(rpart)
library(rpart.plot)
rpart = NULL
# Create a decision tree
m <-rpart(Species~.,data=iris)
#Plot
rpart.plot(m,extra=2,main="Decision Tree - IRIS")

 

1.8 b Decision trees – Python code

from sklearn.datasets import load_iris
from sklearn.tree import DecisionTreeClassifier
from sklearn import tree
from sklearn.model_selection import train_test_split
import graphviz 

iris = load_iris()
X_train, X_test, y_train, y_test = train_test_split(iris.data, iris.target, random_state = 3)
clf = DecisionTreeClassifier().fit(X_train, y_train)

print('Accuracy of Decision Tree classifier on training set: {:.2f}'
     .format(clf.score(X_train, y_train)))
print('Accuracy of Decision Tree classifier on test set: {:.2f}'
     .format(clf.score(X_test, y_test)))

dot_data = tree.export_graphviz(clf, out_file=None, 
                         feature_names=iris.feature_names,  
                         class_names=iris.target_names,  
                         filled=True, rounded=True,  
                         special_characters=True)  
graph = graphviz.Source(dot_data)  
graph
## Accuracy of Decision Tree classifier on training set: 1.00
## Accuracy of Decision Tree classifier on test set: 0.97

1.9a Feature importance – R code

I found the following code which had a snippet for feature importance. Sklean has a nice method for this. For some reason the results in R and Python are different. Any thoughts?

set.seed(3)
# load the library
library(mlbench)
library(caret)
# load the dataset
cancer <- read.csv("cancer.csv")
cancer$target <- as.factor(cancer$target)
# Split as data
data <- cancer[,1:31]
target <- cancer[,32]

# Train the model
model <- train(data, target, method="rf", preProcess="scale", trControl=trainControl(method = "cv"))
# Compute variable importance
importance <- varImp(model)
# summarize importance
print(importance)
# plot importance
plot(importance)

1.9b Feature importance – Python code

import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
from sklearn.tree import DecisionTreeClassifier
from sklearn.model_selection import train_test_split
from sklearn.datasets import load_breast_cancer
import numpy as np
# Read the data
cancer= load_breast_cancer()
(X_cancer, y_cancer) = load_breast_cancer(return_X_y = True)
X_train, X_test, y_train, y_test = train_test_split(X_cancer, y_cancer, random_state = 0)
# Use the DecisionTreClassifier
clf = DecisionTreeClassifier(max_depth = 4, min_samples_leaf = 8,
                            random_state = 0).fit(X_train, y_train)

c_features=len(cancer.feature_names)
print('Breast cancer dataset: decision tree')
print('Accuracy of DT classifier on training set: {:.2f}'
     .format(clf.score(X_train, y_train)))
print('Accuracy of DT classifier on test set: {:.2f}'
     .format(clf.score(X_test, y_test)))

# Plot the feature importances
fig4=plt.figure(figsize=(10,6),dpi=80)

fig4=plt.barh(range(c_features), clf.feature_importances_)
fig4=plt.xlabel("Feature importance")
fig4=plt.ylabel("Feature name")
fig4=plt.yticks(np.arange(c_features), cancer.feature_names)
fig4=plt.tight_layout()
plt.savefig('fig4.png', bbox_inches='tight')
## Breast cancer dataset: decision tree
## Accuracy of DT classifier on training set: 0.96
## Accuracy of DT classifier on test set: 0.94

Output image: 

1.10a Precision-Recall, ROC curves & AUC- R code

I tried several R packages for plotting the Precision and Recall and AUC curve. PRROC seems to work well. The Precision-Recall curves show the tradeoff between precision and recall. The higher the precision, the lower the recall and vice versa.AUC curves that hug the top left corner indicate a high sensitivity,specificity and an excellent accuracy.

source("RFunctions-1.R")
library(dplyr)
library(caret)
library(e1071)
library(PRROC)
# Read the data (this data is from sklearn!)
d <- read.csv("digits.csv")
digits <- d[2:66]
digits$X64 <- as.factor(digits$X64)

# Split as training and test sets
train_idx <- trainTestSplit(digits,trainPercent=75,seed=5)
train <- digits[train_idx, ]
test <- digits[-train_idx, ]

# Fit a SVM model with linear basis kernel with probabilities
svmfit=svm(X64~., data=train, kernel="linear",scale=FALSE,probability=TRUE)
ypred=predict(svmfit,test,probability=TRUE)
head(attr(ypred,"probabilities"))
##               0            1
## 6  7.395947e-01 2.604053e-01
## 8  9.999998e-01 1.842555e-07
## 12 1.655178e-05 9.999834e-01
## 13 9.649997e-01 3.500032e-02
## 15 9.994849e-01 5.150612e-04
## 16 9.999987e-01 1.280700e-06
# Store the probability of 0s and 1s
m0<-attr(ypred,"probabilities")[,1]
m1<-attr(ypred,"probabilities")[,2]

# Create a dataframe of scores
scores <- data.frame(m1,test$X64)

# Class 0 is data points of +ve class (in this case, digit 1) and -ve class (digit 0)
#Compute Precision Recall
pr <- pr.curve(scores.class0=scores[scores$test.X64=="1",]$m1,
               scores.class1=scores[scores$test.X64=="0",]$m1,
               curve=T)

# Plot precision-recall curve
plot(pr)

#Plot the ROC curve
roc<-roc.curve(m0, m1,curve=TRUE)
plot(roc)

1.10b Precision-Recall, ROC curves & AUC- Python code

For Python Logistic Regression is used to plot Precision Recall, ROC curve and compute AUC

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.datasets import load_digits
from sklearn.metrics import precision_recall_curve
from sklearn.metrics import roc_curve, auc
#Load the digits
dataset = load_digits()
X, y = dataset.data, dataset.target
#Create 2 classes -i) Digit 1 (from digit 1) ii) Digit 0 (from all other digits)
# Make a copy of the target
z= y.copy()
# Replace all non 1's as 0
z[z != 1] = 0

X_train, X_test, y_train, y_test = train_test_split(X, z, random_state=0)
# Fit a LR model
lr = LogisticRegression().fit(X_train, y_train)

#Compute the decision scores
y_scores_lr = lr.fit(X_train, y_train).decision_function(X_test)
y_score_list = list(zip(y_test[0:20], y_scores_lr[0:20]))

#Show the decision_function scores for first 20 instances
y_score_list

precision, recall, thresholds = precision_recall_curve(y_test, y_scores_lr)
closest_zero = np.argmin(np.abs(thresholds))
closest_zero_p = precision[closest_zero]
closest_zero_r = recall[closest_zero]
#Plot
plt.figure()
plt.xlim([0.0, 1.01])
plt.ylim([0.0, 1.01])
plt.plot(precision, recall, label='Precision-Recall Curve')
plt.plot(closest_zero_p, closest_zero_r, 'o', markersize = 12, fillstyle = 'none', c='r', mew=3)
plt.xlabel('Precision', fontsize=16)
plt.ylabel('Recall', fontsize=16)
plt.axes().set_aspect('equal')
plt.savefig('fig5.png', bbox_inches='tight')

#Compute and plot the ROC
y_score_lr = lr.fit(X_train, y_train).decision_function(X_test)
fpr_lr, tpr_lr, _ = roc_curve(y_test, y_score_lr)
roc_auc_lr = auc(fpr_lr, tpr_lr)

plt.figure()
plt.xlim([-0.01, 1.00])
plt.ylim([-0.01, 1.01])
plt.plot(fpr_lr, tpr_lr, lw=3, label='LogRegr ROC curve (area = {:0.2f})'.format(roc_auc_lr))
plt.xlabel('False Positive Rate', fontsize=16)
plt.ylabel('True Positive Rate', fontsize=16)
plt.title('ROC curve (1-of-10 digits classifier)', fontsize=16)
plt.legend(loc='lower right', fontsize=13)
plt.plot([0, 1], [0, 1], color='navy', lw=3, linestyle='--')
plt.axes()
plt.savefig('fig6.png', bbox_inches='tight')

output

output

1.10c Precision-Recall, ROC curves & AUC- Python code

In the code below classification probabilities are used to compute and plot precision-recall, roc and AUC

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.datasets import load_digits
from sklearn.svm import LinearSVC
from sklearn.calibration import CalibratedClassifierCV

dataset = load_digits()
X, y = dataset.data, dataset.target
# Make a copy of the target
z= y.copy()
# Replace all non 1's as 0
z[z != 1] = 0


X_train, X_test, y_train, y_test = train_test_split(X, z, random_state=0)
svm = LinearSVC()
# Need to use CalibratedClassifierSVC to redict probabilities for lInearSVC
clf = CalibratedClassifierCV(svm) 
clf.fit(X_train, y_train)
y_proba_lr = clf.predict_proba(X_test)
from sklearn.metrics import precision_recall_curve

precision, recall, thresholds = precision_recall_curve(y_test, y_proba_lr[:,1])
closest_zero = np.argmin(np.abs(thresholds))
closest_zero_p = precision[closest_zero]
closest_zero_r = recall[closest_zero]
#plt.figure(figsize=(15,15),dpi=80)
plt.figure()
plt.xlim([0.0, 1.01])
plt.ylim([0.0, 1.01])
plt.plot(precision, recall, label='Precision-Recall Curve')
plt.plot(closest_zero_p, closest_zero_r, 'o', markersize = 12, fillstyle = 'none', c='r', mew=3)
plt.xlabel('Precision', fontsize=16)
plt.ylabel('Recall', fontsize=16)
plt.axes().set_aspect('equal')
plt.savefig('fig7.png', bbox_inches='tight')

output

Note: As with other posts in this series on ‘Practical Machine Learning with R and Python’,   this post is based on these 2 MOOC courses
1. Statistical Learning, Prof Trevor Hastie & Prof Robert Tibesherani, Online Stanford
2. Applied Machine Learning in Python Prof Kevyn-Collin Thomson, University Of Michigan, Coursera

Conclusion

This 4th part looked at SVMs with linear and radial basis, decision trees, precision-recall tradeoff, ROC curves and AUC.

Stick around for further updates. I’ll be back!
Comments, suggestions and correction are welcome.

Also see
1. A primer on Qubits, Quantum gates and Quantum Operations
2. Dabbling with Wiener filter using OpenCV
3. The mind of a programmer
4. Sea shells on the seashore
5. yorkr pads up for the Twenty20s: Part 1- Analyzing team”s match performance

To see all posts see Index of posts

Analysis of IPL T20 matches with yorkr templates


Introduction

In this post I create RMarkdown templates for end-to-end analysis of IPL T20 matches, that are available on Cricsheet based on my R package yorkr.  With these templates you can convert all IPL data which is in yaml format to R dataframes. Further I create data and the necessary templates for analyzing IPL matches, teams and players. All of these can be accessed at yorkrIPLTemplate.

Check out my 2 books on cricket, a) Cricket analytics with cricketr b) Beaten by sheer pace – Cricket analytics with yorkr, now available in both paperback & kindle versions on Amazon!!! Pick up your copies today!

The templates are

  1. Template for conversion and setup – IPLT20Template.Rmd
  2. Any IPL match – IPLMatchtemplate.Rmd
  3. IPL matches between 2 nations – IPLMatches2TeamTemplate.Rmd
  4. A IPL nations performance against all other IPL nations – IPLAllMatchesAllOppnTemplate.Rmd
  5. Analysis of IPL batsmen and bowlers of all IPL nations – IPLBatsmanBowlerTemplate.Rmd

Besides the templates the repository also includes the converted data for all IPL matches I downloaded from Cricsheet in Dec 2016. So this data is complete till the 2016 IPL season. You can recreate the files as more matches are added to Cricsheet site in IPL 2017 and future seasons. This post contains all the steps needed for detailed analysis of IPL matches, teams and IPL player. This will also be my reference in future if I decide to analyze IPL in future!

There will be 5 folders at the root

  1. IPLdata – Match files as yaml from Cricsheet
  2. IPLMatches – Yaml match files converted to dataframes
  3. IPLMatchesBetween2Teams – All Matches between any 2 IPL teams
  4. allMatchesAllOpposition – An IPL teams’s performance against all other teams
  5. BattingBowlingDetails – Batting and bowling details of all IPL teams
library(yorkr)
library(dplyr)

The first few steps take care of the data setup. This needs to be done before any of the analysis of IPL batsmen, bowlers, any IPL match, matches between any 2 IPL countries or analysis of a teams performance against all other countries

There will be 5 folders at the root

  1. data
  2. IPLMatches
  3. IPLMatchesBetween2Teams
  4. allMatchesAllOpposition
  5. BattingBowlingDetails

The source YAML files will be in IPLData folder

1.Create directory of IPLMatches

Some files may give conversions errors. You could try to debug the problem or just remove it from the IPLdata folder. At most 2-4 file will have conversion problems and I usally remove then from the files to be converted.

Also take a look at my GooglyPlus shiny app which was created after performing the same conversion on the Dec 16 data .

convertAllYaml2RDataframesT20("data","IPLMatches")

2.Save all matches between all combinations of IPL nations

This function will create the set of all matches between each IPL team against every other IPL team. This uses the data that was created in IPLMatches, with the convertAllYaml2RDataframesIPL() function.

setwd("./IPLMatchesBetween2Teams")
saveAllMatchesBetween2IPLTeams("../IPLMatches")

3.Save all matches against all opposition

This will create a consolidated dataframe of all matches played by every IPL playing nation against all other nattions. This also uses the data that was created in IPLMatches, with the convertAllYaml2RDataframesIPL() function.

setwd("../allMatchesAllOpposition")
saveAllMatchesAllOppositionIPLT20("../IPLMatches")

4. Create batting and bowling details for each IPL team

These are the current IPL playing teams. You can add to this vector as newer IPL teams start playing IPL. You will get to know all IPL teams by also look at the directory created above namely allMatchesAllOpposition. This also uses the data that was created in IPLMatches, with the convertAllYaml2RDataframesIPL() function.

setwd("../BattingBowlingDetails")
ipl_teams <- list("Chennai Super Kings","Deccan Chargers", "Delhi Daredevils","Kings XI Punjab", 
              "Kochi Tuskers Kerala","Kolkata Knight Riders","Mumbai Indians","Pune Warriors",
              "Rajasthan Royals","Royal Challengers Bangalore","Sunrisers Hyderabad","Gujarat Lions",
                 "Rising Pune Supergiants")

for(i in seq_along(ipl_teams)){
    print(ipl_teams[i])
    val <- paste(ipl_teams[i],"-details",sep="")
    val <- getTeamBattingDetails(ipl_teams[i],dir="../IPLMatches", save=TRUE)

}

for(i in seq_along(ipl_teams)){
    print(ipl_teams[i])
    val <- paste(ipl_teams[i],"-details",sep="")
    val <- getTeamBowlingDetails(ipl_teams[i],dir="../IPLMatches", save=TRUE)

}

5. Get the list of batsmen for a particular IPL team

The following code is needed for analyzing individual IPL batsmen. In IPL a player could have played in multiple IPL teams.

getBatsmen <- function(df){
    bmen <- df %>% distinct(batsman) 
    bmen <- as.character(bmen$batsman)
    batsmen <- sort(bmen)
}
load("Chennai Super Kings-BattingDetails.RData")
csk_details <- battingDetails
load("Deccan Chargers-BattingDetails.RData")
dc_details <- battingDetails
load("Delhi Daredevils-BattingDetails.RData")
dd_details <- battingDetails
load("Kings XI Punjab-BattingDetails.RData")
kxip_details <- battingDetails
load("Kochi Tuskers Kerala-BattingDetails.RData")
ktk_details <- battingDetails
load("Kolkata Knight Riders-BattingDetails.RData")
kkr_details <- battingDetails
load("Mumbai Indians-BattingDetails.RData")
mi_details <- battingDetails
load("Pune Warriors-BattingDetails.RData")
pw_details <- battingDetails
load("Rajasthan Royals-BattingDetails.RData")
rr_details <- battingDetails
load("Royal Challengers Bangalore-BattingDetails.RData")
rcb_details <- battingDetails
load("Sunrisers Hyderabad-BattingDetails.RData")
sh_details <- battingDetails
load("Gujarat Lions-BattingDetails.RData")
gl_details <- battingDetails
load("Rising Pune Supergiants-BattingDetails.RData")
rps_details <- battingDetails

#Get the batsmen for each IPL team
csk_batsmen <- getBatsmen(csk_details)
dc_batsmen <- getBatsmen(dc_details)
dd_batsmen <- getBatsmen(dd_details)
kxip_batsmen <- getBatsmen(kxip_details)
ktk_batsmen <- getBatsmen(ktk_details)
kkr_batsmen <- getBatsmen(kkr_details)
mi_batsmen <- getBatsmen(mi_details)
pw_batsmen <- getBatsmen(pw_details)
rr_batsmen <- getBatsmen(rr_details)
rcb_batsmen <- getBatsmen(rcb_details)
sh_batsmen <- getBatsmen(sh_details)
gl_batsmen <- getBatsmen(gl_details)
rps_batsmen <- getBatsmen(rps_details)

# Save the dataframes
save(csk_batsmen,file="csk.RData")
save(dc_batsmen, file="dc.RData")
save(dd_batsmen, file="dd.RData")
save(kxip_batsmen, file="kxip.RData")
save(ktk_batsmen, file="ktk.RData")
save(kkr_batsmen, file="kkr.RData")
save(mi_batsmen , file="mi.RData")
save(pw_batsmen, file="pw.RData")
save(rr_batsmen, file="rr.RData")
save(rcb_batsmen, file="rcb.RData")
save(sh_batsmen, file="sh.RData")
save(gl_batsmen, file="gl.RData")
save(rps_batsmen, file="rps.RData")

6. Get the list of bowlers for a particular IPL team

The method below can get the list of bowler names for any IPL team.The following code is needed for analyzing individual IPL bowlers. In IPL a player could have played in multiple IPL teams.

getBowlers <- function(df){
    bwlr <- df %>% distinct(bowler) 
    bwlr <- as.character(bwlr$bowler)
    bowler <- sort(bwlr)
}

load("Chennai Super Kings-BowlingDetails.RData")
csk_details <- bowlingDetails
load("Deccan Chargers-BowlingDetails.RData")
dc_details <- bowlingDetails
load("Delhi Daredevils-BowlingDetails.RData")
dd_details <- bowlingDetails
load("Kings XI Punjab-BowlingDetails.RData")
kxip_details <- bowlingDetails
load("Kochi Tuskers Kerala-BowlingDetails.RData")
ktk_details <- bowlingDetails
load("Kolkata Knight Riders-BowlingDetails.RData")
kkr_details <- bowlingDetails
load("Mumbai Indians-BowlingDetails.RData")
mi_details <- bowlingDetails
load("Pune Warriors-BowlingDetails.RData")
pw_details <- bowlingDetails
load("Rajasthan Royals-BowlingDetails.RData")
rr_details <- bowlingDetails
load("Royal Challengers Bangalore-BowlingDetails.RData")
rcb_details <- bowlingDetails
load("Sunrisers Hyderabad-BowlingDetails.RData")
sh_details <- bowlingDetails
load("Gujarat Lions-BowlingDetails.RData")
gl_details <- bowlingDetails
load("Rising Pune Supergiants-BowlingDetails.RData")
rps_details <- bowlingDetails

# Get the bowlers for each team
csk_bowlers <- getBowlers(csk_details)
dc_bowlers <- getBowlers(dc_details)
dd_bowlers <- getBowlers(dd_details)
kxip_bowlers <- getBowlers(kxip_details)
ktk_bowlers <- getBowlers(ktk_details)
kkr_bowlers <- getBowlers(kkr_details)
mi_bowlers <- getBowlers(mi_details)
pw_bowlers <- getBowlers(pw_details)
rr_bowlers <- getBowlers(rr_details)
rcb_bowlers <- getBowlers(rcb_details)
sh_bowlers <- getBowlers(sh_details)
gl_bowlers <- getBowlers(gl_details)
rps_bowlers <- getBowlers(rps_details)

#Save the dataframes
save(csk_bowlers,file="csk1.RData")
save(dc_bowlers, file="dc1.RData")
save(dd_bowlers, file="dd1.RData")
save(kxip_bowlers, file="kxip1.RData")
save(ktk_bowlers, file="ktk1.RData")
save(kkr_bowlers, file="kkr1.RData")
save(mi_bowlers , file="mi1.RData")
save(pw_bowlers, file="pw1.RData")
save(rr_bowlers, file="rr1.RData")
save(rcb_bowlers, file="rcb1.RData")
save(sh_bowlers, file="sh1.RData")
save(gl_bowlers, file="gl1.RData")
save(rps_bowlers, file="rps1.RData")

Now we are all set

A)  IPL T20 Match Analysis

1 IPL Match Analysis

Load any match data from the ./IPLMatches folder for e.g. Chennai Super Kings-Deccan Chargers-2008-05-06.RData

setwd("./IPLMatches")
load("Chennai Super Kings-Deccan Chargers-2008-05-06.RData")
csk_dc<- overs
#The steps are
load("IPLTeam1-IPLTeam2-Date.Rdata")
IPLTeam1_IPLTeam2 <- overs

All analysis for this match can be done now

2. Scorecard

teamBattingScorecardMatch(IPLTeam1_IPLTeam2,"IPLTeam1")
teamBattingScorecardMatch(IPLTeam1_IPLTeam2,"IPLTeam2")

3.Batting Partnerships

teamBatsmenPartnershipMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")
teamBatsmenPartnershipMatch(IPLTeam1_IPLTeam2,"IPLTeam2","IPLTeam1")

4. Batsmen vs Bowler Plot

teamBatsmenVsBowlersMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=TRUE)
teamBatsmenVsBowlersMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)

5. Team bowling scorecard

teamBowlingScorecardMatch(IPLTeam1_IPLTeam2,"IPLTeam1")
teamBowlingScorecardMatch(IPLTeam1_IPLTeam2,"IPLTeam2")

6. Team bowling Wicket kind match

teamBowlingWicketKindMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")
m <-teamBowlingWicketKindMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)
m

7. Team Bowling Wicket Runs Match

teamBowlingWicketRunsMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")
m <-teamBowlingWicketRunsMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)
m

8. Team Bowling Wicket Match

m <-teamBowlingWicketMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)
m
teamBowlingWicketMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")

9. Team Bowler vs Batsmen

teamBowlersVsBatsmenMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")
m <- teamBowlersVsBatsmenMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)
m

10. Match Worm chart

matchWormGraph(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")

B)  IPL  Matches between 2  IPL teams

1 IPL Match Analysis

Load any match data from the ./IPLMatches folder for e.g. Chennai Super Kings-Deccan Chargers-2008-05-06.RData

setwd("./IPLMatches")
load("Chennai Super Kings-Deccan Chargers-2008-05-06.RData")
csk_dc<- overs
#The steps are
load("IPLTeam1-IPLTeam2-Date.Rdata")
IPLTeam1_IPLTeam2 <- overs

All analysis for this match can be done now

2. Scorecard

teamBattingScorecardMatch(IPLTeam1_IPLTeam2,"IPLTeam1")
teamBattingScorecardMatch(IPLTeam1_IPLTeam2,"IPLTeam2")

3.Batting Partnerships

teamBatsmenPartnershipMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")
teamBatsmenPartnershipMatch(IPLTeam1_IPLTeam2,"IPLTeam2","IPLTeam1")

4. Batsmen vs Bowler Plot

teamBatsmenVsBowlersMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=TRUE)
teamBatsmenVsBowlersMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)

5. Team bowling scorecard

teamBowlingScorecardMatch(IPLTeam1_IPLTeam2,"IPLTeam1")
teamBowlingScorecardMatch(IPLTeam1_IPLTeam2,"IPLTeam2")

6. Team bowling Wicket kind match

teamBowlingWicketKindMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")
m <-teamBowlingWicketKindMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)
m

7. Team Bowling Wicket Runs Match

teamBowlingWicketRunsMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")
m <-teamBowlingWicketRunsMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)
m

8. Team Bowling Wicket Match

m <-teamBowlingWicketMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)
m
teamBowlingWicketMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")

9. Team Bowler vs Batsmen

teamBowlersVsBatsmenMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")
m <- teamBowlersVsBatsmenMatch(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2",plot=FALSE)
m

10. Match Worm chart

matchWormGraph(IPLTeam1_IPLTeam2,"IPLTeam1","IPLTeam2")

C)  IPL Matches for a team against all other teams

1. IPL Matches for a team against all other teams

Load the data between for a IPL team against all other countries ./allMatchesAllOpposition for e.g all matches of Kolkata Knight Riders

load("allMatchesAllOpposition-Kolkata Knight Riders.RData")
kkr_matches <- matches
IPLTeam="IPLTeam1"
allMatches <- paste("allMatchesAllOposition-",IPLTeam,".RData",sep="")
load(allMatches)
IPLTeam1AllMatches <- matches

2. Team’s batting scorecard all Matches

m <-teamBattingScorecardAllOppnAllMatches(IPLTeam1AllMatches,theTeam="IPLTeam1")
m

3. Batting scorecard of opposing team

m <-teamBattingScorecardAllOppnAllMatches(matches=IPLTeam1AllMatches,theTeam="IPLTeam2")

4. Team batting partnerships

m <- teamBatsmenPartnershipAllOppnAllMatches(IPLTeam1AllMatches,theTeam="IPLTeam1")
m
m <- teamBatsmenPartnershipAllOppnAllMatches(IPLTeam1AllMatches,theTeam='IPLTeam1',report="detailed")
head(m,30)
m <- teamBatsmenPartnershipAllOppnAllMatches(IPLTeam1AllMatches,theTeam='IPLTeam1',report="summary")
m

5. Team batting partnerships plot

teamBatsmenPartnershipAllOppnAllMatchesPlot(IPLTeam1AllMatches,"IPLTeam1",main="IPLTeam1")
teamBatsmenPartnershipAllOppnAllMatchesPlot(IPLTeam1AllMatches,"IPLTeam1",main="IPLTeam2")

6, Team batsmen vs bowlers report

m <-teamBatsmenVsBowlersAllOppnAllMatchesRept(IPLTeam1AllMatches,"IPLTeam1",rank=0)
m
m <-teamBatsmenVsBowlersAllOppnAllMatchesRept(IPLTeam1AllMatches,"IPLTeam1",rank=1,dispRows=30)
m
m <-teamBatsmenVsBowlersAllOppnAllMatchesRept(matches=IPLTeam1AllMatches,theTeam="IPLTeam2",rank=1,dispRows=25)
m

7. Team batsmen vs bowler plot

d <- teamBatsmenVsBowlersAllOppnAllMatchesRept(IPLTeam1AllMatches,"IPLTeam1",rank=1,dispRows=50)
d
teamBatsmenVsBowlersAllOppnAllMatchesPlot(d)
d <- teamBatsmenVsBowlersAllOppnAllMatchesRept(IPLTeam1AllMatches,"IPLTeam1",rank=2,dispRows=50)
teamBatsmenVsBowlersAllOppnAllMatchesPlot(d)

8. Team bowling scorecard

teamBowlingScorecardAllOppnAllMatchesMain(matches=IPLTeam1AllMatches,theTeam="IPLTeam1")
teamBowlingScorecardAllOppnAllMatches(IPLTeam1AllMatches,'IPLTeam2')

9. Team bowler vs batsmen

teamBowlersVsBatsmenAllOppnAllMatchesMain(IPLTeam1AllMatches,theTeam="IPLTeam1",rank=0)
teamBowlersVsBatsmenAllOppnAllMatchesMain(IPLTeam1AllMatches,theTeam="IPLTeam1",rank=2)
teamBowlersVsBatsmenAllOppnAllMatchesRept(matches=IPLTeam1AllMatches,theTeam="IPLTeam1",rank=0)

10. Team Bowler vs bastmen

df <- teamBowlersVsBatsmenAllOppnAllMatchesRept(IPLTeam1AllMatches,theTeam="IPLTeam1",rank=1)
teamBowlersVsBatsmenAllOppnAllMatchesPlot(df,"IPLTeam1","IPLTeam1")

11. Team bowler wicket kind

teamBowlingWicketKindAllOppnAllMatches(IPLTeam1AllMatches,t1="IPLTeam1",t2="All")
teamBowlingWicketKindAllOppnAllMatches(IPLTeam1AllMatches,t1="IPLTeam1",t2="IPLTeam2")

12.

teamBowlingWicketRunsAllOppnAllMatches(IPLTeam1AllMatches,t1="IPLTeam1",t2="All",plot=TRUE)
teamBowlingWicketRunsAllOppnAllMatches(IPLTeam1AllMatches,t1="IPLTeam1",t2="IPLTeam2",plot=TRUE)

1 IPL Batsman setup functions

Get the batsman’s details for a batsman

setwd("../BattingBowlingDetails")
# IPL Team names
IPLTeamNames <- list("Chennai Super Kings","Deccan Chargers", "Delhi Daredevils","Kings Xi Punjab", 
                  "Kochi Tuskers Kerala","Kolkata Knight Riders","Mumbai Indians","Pune Warriors",
                  "Rajasthan Royals","Royal Challengers Bangalore","Sunrisers Hyderabad","Gujarat Lions",
                  "Rising Pune Supergiants")           


# Check and get the team indices of IPL teams in which the batsman has played
getTeamIndex <- function(batsman){
    setwd("./BattingBowlingDetails")
    load("csk.RData")
    load("dc.RData")
    load("dd.RData")
    load("kxip.RData")
    load("ktk.RData")
    load("kkr.RData")
    load("mi.RData")
    load("pw.RData")
    load("rr.RData")
    load("rcb.RData")
    load("sh.RData")
    load("gl.RData")
    load("rps.RData")
    setwd("..")
    getwd()
    print(ls())
    teams_batsmen = list(csk_batsmen,dc_batsmen,dd_batsmen,kxip_batsmen,ktk_batsmen,kkr_batsmen,mi_batsmen,
                         pw_batsmen,rr_batsmen,rcb_batsmen,sh_batsmen,gl_batsmen,rps_batsmen)
    b <- NULL
    for (i in 1:length(teams_batsmen)){
        a <- which(teams_batsmen[[i]] == batsman)

        if(length(a) != 0)
            b <- c(b,i)
    }
    b
}

# Get the list of the IPL team names from the indices passed
getTeams <- function(x){

    l <- NULL
    # Get the teams passed in as indexes
    for (i in seq_along(x)){

        l <- c(l, IPLTeamNames[[x[i]]]) 

    }
    l
}

# Create a consolidated data frame with all teams the IPL batsman has played for
getIPLBatsmanDF <- function(teamNames){
    batsmanDF <- NULL
   # Create a consolidated Data frame of batsman for all IPL teams played
    for (i in seq_along(teamNames)){
       df <- getBatsmanDetails(team=teamNames[i],name=IPLBatsman,dir="./BattingBowlingDetails")
       batsmanDF <- rbind(batsmanDF,df) 

    }
    batsmanDF
}

2. Create a consolidated IPL batsman data frame

# Since an IPL batsman coculd have played in multiple teams we need to determine these teams and
# create a consolidated data frame for the analysis
# For example to check MS Dhoni we need to do the following

IPLBatsman = "MS Dhoni"
#Check and get the team indices of IPL teams in which the batsman has played
i <- getTeamIndex(IPLBatsman)

# Get the team names in which the IPL batsman has played
teamNames <- getTeams(i)
    # Check if file exists in the directory. This check is necessary when moving between matchType


############## Create a consolidated IPL batsman dataframe for analysis
batsmanDF <- getIPLBatsmanDF(teamNames)

3. Runs vs deliveries

# For e.g. batsmanName="MS Dhoni""
#batsmanRunsVsDeliveries(batsmanDF, "MS Dhoni")
batsmanRunsVsDeliveries(batsmanDF,"batsmanName")

4. Batsman 4s & 6s

batsman46 <- select(batsmanDF,batsman,ballsPlayed,fours,sixes,runs)
p1 <- batsmanFoursSixes(batsman46,"batsmanName")

5. Batsman dismissals

batsmanDismissals(batsmanDF,"batsmanName")

6. Runs vs Strike rate

batsmanRunsVsStrikeRate(batsmanDF,"batsmanName")

7. Batsman Moving Average

batsmanMovingAverage(batsmanDF,"batsmanName")

8. Batsman cumulative average

batsmanCumulativeAverageRuns(batsmanDF,"batsmanName")

9. Batsman cumulative strike rate

batsmanCumulativeStrikeRate(batsmanDF,"batsmanName")

10. Batsman runs against oppositions

batsmanRunsAgainstOpposition(batsmanDF,"batsmanName")

11. Batsman runs vs venue

batsmanRunsVenue(batsmanDF,"batsmanName")

12. Batsman runs predict

batsmanRunsPredict(batsmanDF,"batsmanName")

13.Bowler set up functions

setwd("../BattingBowlingDetails")
# IPL Team names
IPLTeamNames <- list("Chennai Super Kings","Deccan Chargers", "Delhi Daredevils","Kings Xi Punjab", 
                  "Kochi Tuskers Kerala","Kolkata Knight Riders","Mumbai Indians","Pune Warriors",
                  "Rajasthan Royals","Royal Challengers Bangalore","Sunrisers Hyderabad","Gujarat Lions",
                  "Rising Pune Supergiants")    



# Get the team indices of IPL teams for which the bowler as played
getTeamIndex_bowler <- function(bowler){
    # Load IPL Bowlers
    setwd("./data")
    load("csk1.RData")
    load("dc1.RData")
    load("dd1.RData")
    load("kxip1.RData")
    load("ktk1.RData")
    load("kkr1.RData")
    load("mi1.RData")
    load("pw1.RData")
    load("rr1.RData")
    load("rcb1.RData")
    load("sh1.RData")
    load("gl1.RData")
    load("rps1.RData")
    setwd("..")
    teams_bowlers = list(csk_bowlers,dc_bowlers,dd_bowlers,kxip_bowlers,ktk_bowlers,kkr_bowlers,mi_bowlers,
                         pw_bowlers,rr_bowlers,rcb_bowlers,sh_bowlers,gl_bowlers,rps_bowlers)
    b <- NULL
    for (i in 1:length(teams_bowlers)){
        a <- which(teams_bowlers[[i]] == bowler)
        if(length(a) != 0){
            b <- c(b,i)
        }
    }
    b
}


# Get the list of the IPL team names from the indices passed
getTeams <- function(x){

    l <- NULL
    # Get the teams passed in as indexes
    for (i in seq_along(x)){

        l <- c(l, IPLTeamNames[[x[i]]]) 

    }
    l
}

# Get the team names
teamNames <- getTeams(i)

getIPLBowlerDF <- function(teamNames){
    bowlerDF <- NULL

    # Create a consolidated Data frame of batsman for all IPL teams played
    for (i in seq_along(teamNames)){
          df <- getBowlerWicketDetails(team=teamNames[i],name=IPLBowler,dir="./BattingBowlingDetails")
          bowlerDF <- rbind(bowlerDF,df) 

    }
    bowlerDF
}

14. Get the consolidated data frame for an IPL bowler

# Since an IPL bowler could have played in multiple teams we need to determine these teams and
# create a consolidated data frame for the analysis
# For example to check R Ashwin we need to do the following

IPLBowler = "R Ashwin"
#Check and get the team indices of IPL teams in which the batsman has played
i <- getTeamIndex(IPLBowler)

# Get the team names in which the IPL batsman has played
teamNames <- getTeams(i)
    # Check if file exists in the directory. This check is necessary when moving between matchType


############## Create a consolidated IPL batsman dataframe for analysis
bowlerDF <- getIPLBowlerDF(teamNames)

15. Bowler Mean Economy rate

# For e.g. to get the details of R Ashwin do
#bowlerMeanEconomyRate(bowlerDF,"R Ashwin")
bowlerMeanEconomyRate(bowlerDF,"bowlerName")

16. Bowler mean runs conceded

bowlerMeanRunsConceded(bowlerDF,"bowlerName")

17. Bowler Moving Average

bowlerMovingAverage(bowlerDF,"bowlerName")

18. Bowler cumulative average wickets

bowlerCumulativeAvgWickets(bowlerDF,"bowlerName")

19. Bowler cumulative Economy Rate (ER)

bowlerCumulativeAvgEconRate(bowlerDF,"bowlerName")

20. Bowler wicket plot

bowlerWicketPlot(bowlerDF,"bowlerName")

21. Bowler wicket against opposition

bowlerWicketsAgainstOpposition(bowlerDF,"bowlerName")

22. Bowler wicket at cricket grounds

bowlerWicketsVenue(bowlerDF,"bowlerName")

23. Predict number of deliveries to wickets

setwd("./IPLMatches")
bowlerDF1 <- getDeliveryWickets(team="IPLTeam1",dir=".",name="bowlerName",save=FALSE)
bowlerWktsPredict(bowlerDF1,"bowlerName")

Beaten by sheer pace – Cricket analytics with yorkr


coverMy ebook “Beaten by sheer pace – Cricket analytics with yorkr’  has been published in Leanpub.  You can now download the book (hot off the press!)  for all formats to your favorite device (mobile, iPad, tablet, Kindle)  from the Leanpub  “Beaten by sheer pace!”. The book has been published in the following formats namely

  • PDF (for your computer)
  • EPUB (for iPad or tablets. Save the file cricketAnalyticsWithYorkr.epub to Google Drive/Dropbox and choose “Open in” iBooks for iPad)
  • MOBI (for Kindle. For this format, I suggest that you download & install SendToKindle for PC/Mac. You can then right click the downloaded cricketAnalyticsWithYorkr.mobi and choose SendToKindle. You will need to login to your Kindle account)

From Leanpub
UntitledLeanpub uses a variable pricing model. I have priced the book attractively (I think!). You can choose a price between FREE (limited time offer!) to $4.99 . The link is “Beaten by sheer pace!

This format works with all type Kindle device, Kindle app, Android tablet, iPad.

Checkout my interactive Shiny apps GooglyPlus (plots & tables) and Googly (only plots) which can be used to analyze IPL players, teams and matches.

Introducing cricket package yorkr:Part 4-In the block hole!


Introduction

“The nitrogen in our DNA, the calcium in our teeth, the iron in our blood, the carbon in our apple pies were made in the interiors of collapsing stars. We are made of starstuff.”

“If you wish to make an apple pie from scratch, you must first invent the universe.”

“We are like butterflies who flutter for a day and think it is forever.”

“The absence of evidence is not the evidence of absence.”

“We are star stuff which has taken its destiny into its own hands.”

                              Cosmos - Carl Sagan

This post is the 4th and possibly, the last part of my introduction, to my latest cricket package yorkr. This is the 4th part of the introduction, the 3 earlier ones were

  1. Introducing cricket package yorkr-Part1:Beaten by sheer pace!.
  2. Introducing cricket package yorkr: Part 2-Trapped leg before wicket!
  3. Introducing cricket package yorkr: Part 3-Foxed by flight!

The 1st part included functions dealing with a specific match, the 2nd part dealt with functions between 2 opposing teams. The 3rd part dealt with functions between a team and all matches with all oppositions. This 4th part includes individual batting and bowling performances in ODI matches and deals with Class 4 functions.

This post has also been published at RPubs yorkr-Part4 and can also be downloaded as a PDF document from yorkr-Part4.pdf.

You can clone/fork the code for the package yorkr from Github at yorkr-package

Check out my 2 books on cricket, a) Cricket analytics with cricketr b) Beaten by sheer pace – Cricket analytics with yorkr, now available in both paperback & kindle versions on Amazon!!! Pick up your copies today!

Checkout my interactive Shiny apps GooglyPlus (plots & tables) and Googly (only plots) which can be used to analyze IPL players, teams and matches.

Batsman functions

  1. batsmanRunsVsDeliveries
  2. batsmanFoursSixes
  3. batsmanDismissals
  4. batsmanRunsVsStrikeRate
  5. batsmanMovingAverage
  6. batsmanCumulativeAverageRuns
  7. batsmanCumulativeStrikeRate
  8. batsmanRunsAgainstOpposition
  9. batsmanRunsVenue
  10. batsmanRunsPredict

Bowler functions

  1. bowlerMeanEconomyRate
  2. bowlerMeanRunsConceded
  3. bowlerMovingAverage
  4. bowlerCumulativeAvgWickets
  5. bowlerCumulativeAvgEconRate
  6. bowlerWicketPlot
  7. bowlerWicketsAgainstOpposition
  8. bowlerWicketsVenue
  9. bowlerWktsPredict

Note: The yorkr package in its current avatar only supports ODI, T20 and IPL T20 matches.

library(yorkr)
library(gridExtra)
library(rpart.plot)
library(dplyr)
library(ggplot2)
rm(list=ls())

A. Batsman functions

1. Get Team Batting details

The function below gets the overall team batting details based on the RData file available in ODI matches. This is currently also available in Github at (https://github.com/tvganesh/yorkrData/tree/master/ODI/ODI-matches).  However you may have to do this as future matches are added! The batting details of the team in each match is created and a huge data frame is created by rbinding the individual dataframes. This can be saved as a RData file

setwd("C:/software/cricket-package/york-test/yorkrData/ODI/ODI-matches")
india_details <- getTeamBattingDetails("India",dir=".", save=TRUE)
dim(india_details)
## [1] 11085    15
sa_details <- getTeamBattingDetails("South Africa",dir=".",save=TRUE)
dim(sa_details)
## [1] 6375   15
nz_details <- getTeamBattingDetails("New Zealand",dir=".",save=TRUE)
dim(nz_details)
## [1] 6262   15
eng_details <- getTeamBattingDetails("England",dir=".",save=TRUE)
dim(eng_details)
## [1] 9001   15

2. Get batsman details

This function is used to get the individual batting record for a the specified batsmen of the country as in the functions below. For analyzing the batting performances the following cricketers have been chosen

  1. Virat Kohli (Ind)
  2. M S Dhoni (Ind)
  3. AB De Villiers (SA)
  4. Q De Kock (SA)
  5. J Root (Eng)
  6. M J Guptill (NZ)
setwd("C:/software/cricket-package/york-test/yorkrData/ODI/ODI-matches")
kohli <- getBatsmanDetails(team="India",name="Kohli",dir=".")
## [1] "./India-BattingDetails.RData"
dhoni <- getBatsmanDetails(team="India",name="Dhoni")
## [1] "./India-BattingDetails.RData"
devilliers <-  getBatsmanDetails(team="South Africa",name="Villiers",dir=".")
## [1] "./South Africa-BattingDetails.RData"
deKock <-  getBatsmanDetails(team="South Africa",name="Kock",dir=".")
## [1] "./South Africa-BattingDetails.RData"
root <-  getBatsmanDetails(team="England",name="Root",dir=".")
## [1] "./England-BattingDetails.RData"
guptill <-  getBatsmanDetails(team="New Zealand",name="Guptill",dir=".")
## [1] "./New Zealand-BattingDetails.RData"

3. Runs versus deliveries

Kohli, De Villiers and Guptill have a good cluster of points that head towards 150 runs at 150 deliveries.

p1 <-batsmanRunsVsDeliveries(kohli,"Kohli")
p2 <- batsmanRunsVsDeliveries(dhoni, "Dhoni")
p3 <- batsmanRunsVsDeliveries(devilliers,"De Villiers")
p4 <- batsmanRunsVsDeliveries(deKock,"Q de Kock")
p5 <- batsmanRunsVsDeliveries(root,"JE Root")
p6 <- batsmanRunsVsDeliveries(guptill,"MJ Guptill")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

runsVsDeliveries-1

4. Batsman Total runs, Fours and Sixes

The plots below show the total runs, fours and sixes by the batsmen

kohli46 <- select(kohli,batsman,ballsPlayed,fours,sixes,runs)
p1 <- batsmanFoursSixes(kohli46,"Kohli")
dhoni46 <- select(dhoni,batsman,ballsPlayed,fours,sixes,runs)
p2 <- batsmanFoursSixes(dhoni46,"Dhoni")
devilliers46 <- select(devilliers,batsman,ballsPlayed,fours,sixes,runs)
p3 <- batsmanFoursSixes(devilliers46, "De Villiers")
deKock46 <- select(deKock,batsman,ballsPlayed,fours,sixes,runs)
p4 <- batsmanFoursSixes(deKock46,"Q de Kock")
root46 <- select(root,batsman,ballsPlayed,fours,sixes,runs)
p5 <- batsmanFoursSixes(root46,"JE Root")
guptill46 <- select(guptill,batsman,ballsPlayed,fours,sixes,runs)
p6 <- batsmanFoursSixes(guptill46,"MJ Guptill")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

foursSixes-1

5. Batsman dismissals

The type of dismissal for each batsman is shown below

p1 <-batsmanDismissals(kohli,"Kohli")
p2 <- batsmanDismissals(dhoni, "Dhoni")
p3 <- batsmanDismissals(devilliers, "De Villiers")
p4 <- batsmanDismissals(deKock,"Q de Kock")
p5 <- batsmanDismissals(root,"JE Root")
p6 <- batsmanDismissals(guptill,"MJ Guptill")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

dismissal-1

6. Runs versus Strike Rate

De villiers has the best strike rate among all as there are more points to the right side of the plot for the same runs. Kohli and Dhoni do well too. Q De Kock and Joe Root also have a very good spread of points though they have fewer innings.

p1 <-batsmanRunsVsStrikeRate(kohli,"Kohli")
p2 <- batsmanRunsVsStrikeRate(dhoni, "Dhoni")
p3 <- batsmanRunsVsStrikeRate(devilliers, "De Villiers")
p4 <- batsmanRunsVsStrikeRate(deKock,"Q de Kock")
p5 <- batsmanRunsVsStrikeRate(root,"JE Root")
p6 <- batsmanRunsVsStrikeRate(guptill,"MJ Guptill")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

runsSR-1

7. Batsman moving average

Kohli’s average is on a gentle increase from below 50 to around 60’s. Joe Root performance is impressive with his moving average of late tending towards the 70’s. Q De Kock seemed to have a slump around 2015 but his performance is on the increase. Devilliers consistently averages around 50. Dhoni also has been having a stable run in the last several years.

p1 <-batsmanMovingAverage(kohli,"Kohli")
p2 <- batsmanMovingAverage(dhoni, "Dhoni")
p3 <- batsmanMovingAverage(devilliers, "De Villiers")
p4 <- batsmanMovingAverage(deKock,"Q de Kock")
p5 <- batsmanMovingAverage(root,"JE Root")
p6 <- batsmanMovingAverage(guptill,"MJ Guptill")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

ma-1

8. Batsman cumulative average

The functions below provide the cumulative average of runs scored. As can be seen Kohli and Devilliers have a cumulative runs rate that averages around 48-50. Q De Kock seems to have had a rocky career with several highs and lows as the cumulative average oscillates between 45-40. Root steadily improves to a cumulative average of around 42-43 from his 50th innings

p1 <-batsmanCumulativeAverageRuns(kohli,"Kohli")
p2 <- batsmanCumulativeAverageRuns(dhoni, "Dhoni")
p3 <- batsmanCumulativeAverageRuns(devilliers, "De Villiers")
p4 <- batsmanCumulativeAverageRuns(deKock,"Q de Kock")
p5 <- batsmanCumulativeAverageRuns(root,"JE Root")
p6 <- batsmanCumulativeAverageRuns(guptill,"MJ Guptill")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

cAvg-1

9. Cumulative Average Strike Rate

The plots below show the cumulative average strike rate of the batsmen. Dhoni and Devilliers have the best cumulative average strike rate of 90%. The rest average around 80% strike rate. Guptill shows a slump towards the latter part of his career.

p1 <-batsmanCumulativeStrikeRate(kohli,"Kohli")
p2 <- batsmanCumulativeStrikeRate(dhoni, "Dhoni")
p3 <- batsmanCumulativeStrikeRate(devilliers, "De Villiers")
p4 <- batsmanCumulativeStrikeRate(deKock,"Q de Kock")
p5 <- batsmanCumulativeStrikeRate(root,"JE Root")
p6 <- batsmanCumulativeStrikeRate(guptill,"MJ Guptill")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

cSR-1

10. Batsman runs against opposition

Kohli’s best performances are against Australia, West Indies and Sri Lanka

batsmanRunsAgainstOpposition(kohli,"Kohli")

runsOppn1-1

batsmanRunsAgainstOpposition(dhoni, "Dhoni")

runsOppn2-1

Kohli’s best performances are against Australia, Pakistan and West Indies

batsmanRunsAgainstOpposition(devilliers, "De Villiers")

runsOppn3-1

Quentin de Kock average almost 100 runs against India and 75 runs against England

batsmanRunsAgainstOpposition(deKock, "Q de Kock")

runsOppn4-1

Root’s best performances are against South Africa, Sri Lanka and West Indies

batsmanRunsAgainstOpposition(root, "JE Root")

runsOppn5-1

batsmanRunsAgainstOpposition(guptill, "MJ Guptill")

runsOppn6-1

11. Runs at different venues

The plots below give the performances of the batsmen at different grounds.

batsmanRunsVenue(kohli,"Kohli")

runsVenue1-1

batsmanRunsVenue(dhoni, "Dhoni")

runsVenue2-1

batsmanRunsVenue(devilliers, "De Villiers")

runsVenue3-1

batsmanRunsVenue(deKock, "Q de Kock")

runsVenue4-1

batsmanRunsVenue(root, "JE Root")

runsVenue5-1

batsmanRunsVenue(guptill, "MJ Guptill")

runsVenue6-1

12. Predict number of runs to deliveries

The plots below use rpart classification tree to predict the number of deliveries required to score the runs in the leaf node. For e.g. Kohli takes 66 deliveries to score 64 runs and for higher number of deliveries scores around 115 runs. Devilliers needs

par(mfrow=c(1,3))
par(mar=c(4,4,2,2))
batsmanRunsPredict(kohli,"Kohli")
batsmanRunsPredict(dhoni, "Dhoni")
batsmanRunsPredict(devilliers, "De Villiers")

runsPredict1,runsVenue1-1

par(mfrow=c(1,3))
par(mar=c(4,4,2,2))
batsmanRunsPredict(deKock,"Q de Kock")
batsmanRunsPredict(root,"JE Root")
batsmanRunsPredict(guptill,"MJ Guptill")

runsPredict2,runsVenue1-1

B. Bowler functions

13. Get bowling details

The function below gets the overall team bowling details based on the RData file available in ODI matches. This is currently also available in Github at (https://github.com/tvganesh/yorkrData/tree/master/ODI/ODI-matches). The bowling details of the team in each match is created and a huge data frame is created by rbinding the individual dataframes. This can be saved as a RData file

setwd("C:/software/cricket-package/york-test/yorkrData/ODI/ODI-matches")
ind_bowling <- getTeamBowlingDetails("India",dir=".",save=TRUE)
dim(ind_bowling)
## [1] 7816   12
aus_bowling <- getTeamBowlingDetails("Australia",dir=".",save=TRUE)
dim(aus_bowling)
## [1] 9191   12
ban_bowling <- getTeamBowlingDetails("Bangladesh",dir=".",save=TRUE)
dim(ban_bowling)
## [1] 5665   12
sa_bowling <- getTeamBowlingDetails("South Africa",dir=".",save=TRUE)
dim(sa_bowling)
## [1] 3806   12
sl_bowling <- getTeamBowlingDetails("Sri Lanka",dir=".",save=TRUE)
dim(sl_bowling)
## [1] 3964   12

14. Get bowling details of the individual bowlers

This function is used to get the individual bowling record for a specified bowler of the country as in the functions below. For analyzing the bowling performances the following cricketers have been chosen

  1. R A Jadeja (Ind)
  2. Ravichander Ashwin (Ind)
  3. Mitchell Starc (Aus)
  4. Shakib Al Hasan (Ban)
  5. Ajantha Mendis (SL)
  6. Dale Steyn (SA)
jadeja <- getBowlerWicketDetails(team="India",name="Jadeja",dir=".")
ashwin <- getBowlerWicketDetails(team="India",name="Ashwin",dir=".")
starc <-  getBowlerWicketDetails(team="Australia",name="Starc",dir=".")
shakib <-  getBowlerWicketDetails(team="Bangladesh",name="Shakib",dir=".")
mendis <-  getBowlerWicketDetails(team="Sri Lanka",name="Mendis",dir=".")
steyn <-  getBowlerWicketDetails(team="South Africa",name="Steyn",dir=".")

15. Bowler Mean Economy Rate

Shakib Al Hassan is expensive in the 1st 3 overs after which he is very economical with a economy rate of 3-4. Starc, Steyn average around a ER of 4.0

p1<-bowlerMeanEconomyRate(jadeja,"RA Jadeja")
p2<-bowlerMeanEconomyRate(ashwin, "R Ashwin")
p3<-bowlerMeanEconomyRate(starc, "MA Starc")
p4<-bowlerMeanEconomyRate(shakib, "Shakib Al Hasan")
p5<-bowlerMeanEconomyRate(mendis, "A Mendis")
p6<-bowlerMeanEconomyRate(steyn, "D Steyn")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

meanER-1

16. Bowler Mean Runs conceded

Ashwin is expensive around 6 & 7 overs

p1<-bowlerMeanRunsConceded(jadeja,"RA Jadeja")
p2<-bowlerMeanRunsConceded(ashwin, "R Ashwin")
p3<-bowlerMeanRunsConceded(starc, "M A Starc")
p4<-bowlerMeanRunsConceded(shakib, "Shakib Al Hasan")
p5<-bowlerMeanRunsConceded(mendis, "A Mendis")
p6<-bowlerMeanRunsConceded(steyn, "D Steyn")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

meanRunsConceded-1

17. Bowler Moving average

RA jadeja and Mendis’ performance has dipped considerably, while Ashwin and Shakib have improving performances. Starc average around 4 wickets

p1<-bowlerMovingAverage(jadeja,"RA Jadeja")
p2<-bowlerMovingAverage(ashwin, "Ashwin")
p3<-bowlerMovingAverage(starc, "M A Starc")
p4<-bowlerMovingAverage(shakib, "Shakib Al Hasan")
p5<-bowlerMovingAverage(mendis, "Ajantha Mendis")
p6<-bowlerMovingAverage(steyn, "Dale Steyn")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

bowlerMA-1

17. Bowler cumulative average wickets

Starc is clearly the most consistent performer with 3 wickets on an average over his career, while Jadeja averages around 2.0. Ashwin seems to have dropped from 2.4-2.0 wickets, while Mendis drops from high 3.5 to 2.2 wickets. The fractional wickets only show a tendency to take another wicket.

p1<-bowlerCumulativeAvgWickets(jadeja,"RA Jadeja")
p2<-bowlerCumulativeAvgWickets(ashwin, "Ashwin")
p3<-bowlerCumulativeAvgWickets(starc, "M A Starc")
p4<-bowlerCumulativeAvgWickets(shakib, "Shakib Al Hasan")
p5<-bowlerCumulativeAvgWickets(mendis, "Ajantha Mendis")
p6<-bowlerCumulativeAvgWickets(steyn, "Dale Steyn")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

cumWkts-1

18. Bowler cumulative Economy Rate (ER)

The plots below are interesting. All of the bowlers seem to average around 4.5 runs/over. RA Jadeja’s ER improves and heads to 4.5, Mendis is seen to getting more expensive as his career progresses. From a ER of 3.0 he increases towards 4.5

p1<-bowlerCumulativeAvgEconRate(jadeja,"RA Jadeja")
p2<-bowlerCumulativeAvgEconRate(ashwin, "Ashwin")
p3<-bowlerCumulativeAvgEconRate(starc, "M A Starc")
p4<-bowlerCumulativeAvgEconRate(shakib, "Shakib Al Hasan")
p5<-bowlerCumulativeAvgEconRate(mendis, "Ajantha Mendis")
p6<-bowlerCumulativeAvgEconRate(steyn, "Dale Steyn")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

cumER-1

19. Bowler wicket plot

The plot below gives the average wickets versus number of overs

p1<-bowlerWicketPlot(jadeja,"RA Jadeja")
p2<-bowlerWicketPlot(ashwin, "Ashwin")
p3<-bowlerWicketPlot(starc, "M A Starc")
p4<-bowlerWicketPlot(shakib, "Shakib Al Hasan")
p5<-bowlerWicketPlot(mendis, "Ajantha Mendis")
p6<-bowlerWicketPlot(steyn, "Dale Steyn")
grid.arrange(p1,p2,p3,p4,p5,p6, ncol=3)

wktPlot-1

20. Bowler wicket against opposition

#Jadeja's' best pertformance are against England, Pakistan and West Indies
bowlerWicketsAgainstOpposition(jadeja,"RA Jadeja")

wktsOppn1-1

#Ashwin's bets pertformance are against England, Pakistan and South Africa
bowlerWicketsAgainstOpposition(ashwin, "Ashwin")

wktsOppn2-1

#Starc has good performances against India, New Zealand, Pakistan, West Indies
bowlerWicketsAgainstOpposition(starc, "M A Starc")

wktsOppn3-1

bowlerWicketsAgainstOpposition(shakib,"Shakib Al Hasan")

wktsOppn4-1

bowlerWicketsAgainstOpposition(mendis, "Ajantha Mendis")

wktsOppn5-1

#Steyn has good performances against India, Sri Lanka, Pakistan, West Indies
bowlerWicketsAgainstOpposition(steyn, "Dale Steyn")

wktsOppn6-1

21. Bowler wicket at cricket grounds

bowlerWicketsVenue(jadeja,"RA Jadeja")

wktsAve1-1

bowlerWicketsVenue(ashwin, "Ashwin")

wktsAve2-1

bowlerWicketsVenue(starc, "M A Starc")
## Warning: Removed 2 rows containing missing values (geom_bar).

wktsAve3-1

bowlerWicketsVenue(shakib,"Shakib Al Hasan")

wktsAve4-1

bowlerWicketsVenue(mendis, "Ajantha Mendis")

wktsAve5-1

bowlerWicketsVenue(steyn, "Dale Steyn")

wktsAve6-1

22. Get Delivery wickets for bowlers

Thsi function creates a dataframe of deliveries and the wickets taken

setwd("C:/software/cricket-package/york-test/yorkrData/ODI/ODI-matches")
jadeja1 <- getDeliveryWickets(team="India",dir=".",name="Jadeja",save=FALSE)
ashwin1 <- getDeliveryWickets(team="India",dir=".",name="Ashwin",save=FALSE)
starc1 <- getDeliveryWickets(team="Australia",dir=".",name="MA Starc",save=FALSE)
shakib1 <- getDeliveryWickets(team="Bangladesh",dir=".",name="Shakib",save=FALSE)
mendis1 <- getDeliveryWickets(team="Sri Lanka",dir=".",name="Mendis",save=FALSE)
steyn1 <- getDeliveryWickets(team="South Africa",dir=".",name="Steyn",save=FALSE)

23. Predict number of deliveries to wickets

#Jadeja and Ashwin need around 22 to 28 deliveries to make a break through
par(mfrow=c(1,2))
par(mar=c(4,4,2,2))
bowlerWktsPredict(jadeja1,"RA Jadeja")
bowlerWktsPredict(ashwin1,"RAshwin")

wktsPred1-1

#Starc and Shakib provide an early breakthrough producing a wicket in around 16 balls. Starc's 2nd wicket comed around the 30th delivery
par(mfrow=c(1,2))
par(mar=c(4,4,2,2))
bowlerWktsPredict(starc1,"MA Starc")
bowlerWktsPredict(shakib1,"Shakib Al Hasan")

wktsPred2-1

#Steyn and Mendis take 20 deliveries to get their 1st wicket
par(mfrow=c(1,2))
par(mar=c(4,4,2,2))
bowlerWktsPredict(mendis1,"A Mendis")
bowlerWktsPredict(steyn1,"DSteyn")

wktsPred3-1

Conclusion

This concludes the 4 part introduction to my new R cricket package yorkr for ODIs. I will be enhancing the package to handle Twenty20 and IPL matches soon. You can fork/clone the code from Github at yorkr.

The yaml data from Cricsheet have already beeen converted into R consumable dataframes. The converted data can be downloaded from Github at yorkrData. There are 3 folders – ODI matches, ODI matches between 2 teams (oppnAllMatches), ODI matches between a team and the rest of the world (all matches,all oppositions).

As I have already mentioned I have around 67 functions for analysis, however I am certain that the data has a lot more secrets waiting to be tapped. So please do go ahead and run any machine learning or statistical learning algorithms on them. If you do come up with interesting insights, I would appreciate if attribute the source to Cricsheet(http://cricsheet.org), and my package yorkr and my blog Giga thoughts*, besides dropping me a note.

Hope you have a great time with my yorkr package!

Also see

  1. Introducing cricketr! : An R package to analyze performances of cricketers
  2. Cricket analytics with cricketr in paperback and Kindle versions
  3. My TEDx talk on the “Internet of Things”
  4. Bend it like Bluemix,MongoDB with autoscaling – Part 1
  5. The mind of a programmer
  6. Fun simulation of a chain in Android
  7. Taking cricketr for a spin-Part 1
  8. Latency,throughput implications for the cloud
  9. Hand detection through haar-training: A hands-on approach
  10. Cricket analytics with cricketr

Introducing cricket package yorkr: Part 2-Trapped leg before wicket!


“It was a puzzling thing. The truth knocks on the door and you say ‘Go away, I ’m looking for the truth,’ and so it goes away. Puzzling.”

“But even though Quality cannot be defined, you know what Quality is!”

“The Buddha, the Godhead, resides quite comfortably in the circuits of a digital computer or the gears of a cycle transmission as he does at the top of a mountain or in the petals of the flower. To think otherwise is to demean the Buddha – which is to demean oneself.”

                Zen and the Art of Motorcycle maintenance - Robert M Pirsig

Introduction

If we were to to extend the last quote from Zen and the Art of Motorcycle Maintenance, by Robert M Pirsig, I think it would be fair to say that the Buddha also comfortably resides in the exquisite backhand cross-court return of Bjorn Borg, to the the graceful arc of the football in a Lionel Messi’s free kick to the smashing cover drive of Sunil Gavaskar.

In this post I continue to introduce my latest cricket package yorkr. This post is a continuation of my earlier post – Introducing cricket package yorkr-Part1:Beaten by sheer pace!. This post deals with Class 2 functions namely the performances of a team in all matches against a single opposition for e.g all matches of India-Australia, Pakistan-West Indies etc. You can clone/fork the code for my package yorkr from Github at yorkr

Check out my 2 books on cricket, a) Cricket analytics with cricketr b) Beaten by sheer pace – Cricket analytics with yorkr, now available in both paperback & kindle versions on Amazon!!! Pick up your copies today!

Note 1: The package currently only supports ODI, T20s and IPL T20 matches.

This post has also been published at RPubs yorkr-Part2 and can also be downloaded as a PDF document from yorkr-Part2.pdf

Checkout my interactive Shiny apps GooglyPlus (plots & tables) and Googly (only plots) which can be used to analyze IPL players, teams and matches.

The list of function in Class 2 are

  1. teamBatsmenPartnershiOppnAllMatches()
  2. teamBatsmenPartnershipOppnAllMatchesChart()
  3. teamBatsmenVsBowlersOppnAllMatches()
  4. teamBattingScorecardOppnAllMatches()
  5. teamBowlingPerfOppnAllMatches()
  6. teamBowlersWicketsOppnAllMatches()
  7. teamBowlersVsBatsmenOppnAllMatches()
  8. teamBowlersWicketKindOppnAllMatches()
  9. teamBowlersWicketRunsOppnAllMatches()
  10. plotWinLossBetweenTeams()

1. Install the package from CRAN

if (!require("yorkr")) {
  install.packages("yorkr") 
  library("yorkr")
}
library(plotly) 
rm(list=ls())

2. Get data for all matches between 2 teams

We can get all matches between any 2 teams using the function below. The dir parameter should point to the folder which RData files of the individual matches. This function creates a data frame of all the matches and also saves the dataframe as RData

setwd("C:/software/cricket-package/york-test/yorkrData/ODI/ODI-matches")
matches <- getAllMatchesBetweenTeams("Australia","India",dir=".")
dim(matches)
## [1] 67428    25

I have however already saved the matches for all possible combination of opposing countries. The data for these matches for the individual teams/countries can be obtained from Github at in the folder ODI-allmatches-between-two-teams

Note: The dataframe for the different head-to-head matches can be loaded directly into your code. The datframes are 15000+ rows x 25 columns. While I have 10 functions to process the details between teams, feel free to let loose any statistical or machine learning algorithms on the dataframe. So go ahead with any insights that can be gleaned from random forests, ridge regression,SVM classifiers and so on. If you do come up with something interesting, I would appreciate if you could drop me a note. Also please do attribute source to Cricsheet (http://cricsheet.org), the package york and my blog Giga thoughts

3. Save data for all matches between all combination of 2 teams

This can be done locally using the function below. You could use this function to combine all matches between any 2 teams into a single dataframe and save it in the current folder. The current implementation expectes that the the RData files of individual matches are in ../data folder. Since I already have converted this I will not be running this again

#saveAllMatchesBetweenTeams()

4. Load data directly for all matches between 2 teams

As in my earlier post I pick all matches between 2 random teams. I load the data directly from the stored RData files. When we load the Rdata file a “matches” object will be created. This object can be stored for the apporpriate teams as below

setwd("C:/software/cricket-package/york-test/yorkrData/ODI/ODI-allmatches-between-two-teams")
load("India-Australia-allMatches.RData")
aus_ind_matches <- matches
dim(aus_ind_matches)
## [1] 21909    25
load("England-New Zealand-allMatches.RData")
eng_nz_matches <- matches
dim(eng_nz_matches)
## [1] 15343    25
load("Pakistan-South Africa-allMatches.RData")
pak_sa_matches <- matches
dim(pak_sa_matches)
## [1] 17083    25
load("Sri Lanka-West Indies-allMatches.RData")
sl_wi_matches <- matches
dim(sl_wi_matches)
## [1] 4869   25
load("Bangladesh-Ireland-allMatches.RData")
ban_ire_matches <-matches
dim(ban_ire_matches)
## [1] 1668   25
load("Kenya-Bermuda-allMatches.RData")
ken_ber_matches <- matches
dim(ken_ber_matches)
## [1] 1518   25
load("Scotland-Canada-allMatches.RData")
sco_can_matches <-matches
dim(sco_can_matches)
## [1] 1061   25
load("Netherlands-Afghanistan-allMatches.RData")
nl_afg_matches <- matches
dim(nl_afg_matches)
## [1] 402  25

5. Team Batsmen partnership (all matches with opposition)

This function will create a report of the batting partnerships in the teams. The report can be brief or detailed depending on the parameter ‘report’. The top batsmen in India-Australia clashes are Ricky Ponting from Australia and Mahendra Singh Dhoni of India.

m<- teamBatsmenPartnershiOppnAllMatches(aus_ind_matches,'Australia',report="summary")
m
## Source: local data frame [47 x 2]
## 
##       batsman totalRuns
##        (fctr)     (dbl)
## 1  RT Ponting       876
## 2  MEK Hussey       753
## 3   GJ Bailey       614
## 4   SR Watson       609
## 5   MJ Clarke       607
## 6   ML Hayden       573
## 7   A Symonds       536
## 8    AJ Finch       525
## 9   SPD Smith       467
## 10  DA Warner       391
## ..        ...       ...
m <-teamBatsmenPartnershiOppnAllMatches(aus_ind_matches,'India',report="summary")
m
## Source: local data frame [44 x 2]
## 
##         batsman totalRuns
##          (fctr)     (dbl)
## 1      MS Dhoni      1156
## 2     RG Sharma       918
## 3  SR Tendulkar       910
## 4       V Kohli       902
## 5     G Gambhir       536
## 6  Yuvraj Singh       524
## 7      SK Raina       509
## 8      S Dhawan       471
## 9      V Sehwag       289
## 10   RV Uthappa       283
## ..          ...       ...
m <-teamBatsmenPartnershiOppnAllMatches(aus_ind_matches,'Australia',report="detailed")
m <-teamBatsmenPartnershiOppnAllMatches(pak_sa_matches,'Pakistan',report="summary")
m
## Source: local data frame [40 x 2]
## 
##            batsman totalRuns
##             (fctr)     (dbl)
## 1    Misbah-ul-Haq       727
## 2      Younis Khan       657
## 3    Shahid Afridi       558
## 4  Mohammad Yousuf       539
## 5  Mohammad Hafeez       477
## 6     Shoaib Malik       452
## 7    Ahmed Shehzad       348
## 8     Abdul Razzaq       246
## 9     Kamran Akmal       241
## 10      Umar Akmal       215
## ..             ...       ...
m <-teamBatsmenPartnershiOppnAllMatches(eng_nz_matches,'England',report="summary")
m
## Source: local data frame [47 x 2]
## 
##           batsman totalRuns
##            (fctr)     (dbl)
## 1         IR Bell       654
## 2         JE Root       612
## 3  PD Collingwood       514
## 4      EJG Morgan       479
## 5         AN Cook       464
## 6       IJL Trott       362
## 7    KP Pietersen       358
## 8      JC Buttler       287
## 9         OA Shah       274
## 10      RS Bopara       222
## ..            ...       ...
m <-teamBatsmenPartnershiOppnAllMatches(sl_wi_matches,'Sri Lanka',report="summary")
m[1:50,]
## Source: local data frame [50 x 2]
## 
##             batsman totalRuns
##              (fctr)     (dbl)
## 1  DPMD Jayawardene       288
## 2     KC Sangakkara       238
## 3        TM Dilshan       224
## 4       WU Tharanga       220
## 5        AD Mathews       161
## 6     ST Jayasuriya       160
## 7       ML Udawatte        87
## 8   HDRL Thirimanne        67
## 9       MDKJ Perera        64
## 10    CK Kapugedera        57
## ..              ...       ...
m <- teamBatsmenPartnershiOppnAllMatches(ban_ire_matches,"Ireland",report="summary")
m
## Source: local data frame [16 x 2]
## 
##             batsman totalRuns
##              (fctr)     (dbl)
## 1   WTS Porterfield       111
## 2        KJ O'Brien        99
## 3        NJ O'Brien        75
## 4         GC Wilson        60
## 5          AR White        38
## 6       DT Johnston        36
## 7           JP Bray        31
## 8         JF Mooney        28
## 9          AC Botha        23
## 10         EC Joyce        16
## 11      PR Stirling        15
## 12      GH Dockrell         9
## 13        WB Rankin         9
## 14 D Langford-Smith         6
## 15       EJG Morgan         5
## 16        AR Cusack         0

6. Team batsmen partnership (all matches with opposition)

This is plotted graphically in the charts below

teamBatsmenPartnershipOppnAllMatchesChart(aus_ind_matches,"India","Australia")

teamBatsmenPartnership-1

teamBatsmenPartnershipOppnAllMatchesChart(pak_sa_matches,main="South Africa",opposition="Pakistan")

teamBatsmenPartnership-2

m<- teamBatsmenPartnershipOppnAllMatchesChart(eng_nz_matches,"New Zealand",opposition="England",plot=FALSE)
m[1:30,]
##          batsman    nonStriker runs
## 1  KS Williamson   LRPL Taylor  354
## 2    BB McCullum    MJ Guptill  275
## 3    LRPL Taylor KS Williamson  273
## 4     MJ Guptill   BB McCullum  227
## 5    BB McCullum      JD Ryder  212
## 6     MJ Guptill KS Williamson  196
## 7  KS Williamson    MJ Guptill  179
## 8       JD Ryder   BB McCullum  175
## 9       JDP Oram     SB Styris  153
## 10   LRPL Taylor    GD Elliott  147
## 11    GD Elliott   LRPL Taylor  143
## 12   LRPL Taylor    MJ Guptill  140
## 13        JM How   BB McCullum  128
## 14    MJ Guptill   LRPL Taylor  125
## 15   BB McCullum        JM How  117
## 16   BB McCullum   LRPL Taylor  116
## 17     SB Styris      JDP Oram  100
## 18   LRPL Taylor        JM How   98
## 19        JM How   LRPL Taylor   98
## 20      JDP Oram   BB McCullum   84
## 21   LRPL Taylor     L Vincent   71
## 22      JDP Oram    DL Vettori   70
## 23   LRPL Taylor   BB McCullum   61
## 24     SB Styris        JM How   55
## 25      DR Flynn     SB Styris   54
## 26    DL Vettori      JDP Oram   53
## 27     L Vincent   LRPL Taylor   53
## 28    MJ Santner   LRPL Taylor   53
## 29    SP Fleming     L Vincent   52
## 30        JM How     SB Styris   50
teamBatsmenPartnershipOppnAllMatchesChart(sl_wi_matches,"Sri Lanka","West Indies")

teamBatsmenPartnership-3

teamBatsmenPartnershipOppnAllMatchesChart(ban_ire_matches,"Bangladesh","Ireland")

teamBatsmenPartnership-4

7. Team batsmen versus bowler (all matches with opposition)

The plots below provide information on how each of the top batsmen fared against the opposition bowlers

teamBatsmenVsBowlersOppnAllMatches(aus_ind_matches,"India","Australia")

batsmenvsBowler-1

teamBatsmenVsBowlersOppnAllMatches(pak_sa_matches,"South Africa","Pakistan",top=3)

batsmenvsBowler-2

m <- teamBatsmenVsBowlersOppnAllMatches(eng_nz_matches,"England","New Zealnd",top=10,plot=FALSE)
m
## Source: local data frame [157 x 3]
## Groups: batsman [1]
## 
##    batsman       bowler  runs
##     (fctr)       (fctr) (dbl)
## 1  IR Bell JEC Franklin    63
## 2  IR Bell      SE Bond    13
## 3  IR Bell MR Gillespie    33
## 4  IR Bell     NJ Astle     0
## 5  IR Bell     JS Patel    20
## 6  IR Bell   DL Vettori    28
## 7  IR Bell     JDP Oram    48
## 8  IR Bell    SB Styris    12
## 9  IR Bell     KD Mills   124
## 10 IR Bell   TG Southee    84
## ..     ...          ...   ...
teamBatsmenVsBowlersOppnAllMatches(sl_wi_matches,"Sri Lanka","West Indies")

batsmenvsBowler-3

teamBatsmenVsBowlersOppnAllMatches(ban_ire_matches,"Bangladesh","Ireland")

batsmenvsBowler-4

8. Team batsmen versus bowler (all matches with opposition)

The following tables gives the overall performances of the country’s batsmen against the opposition. For India-Australia matches Dhoni, Rohit Sharma and Tendulkar lead the way. For Australia it is Ricky Ponting, M Hussey and GJ Bailey. In South Africa- Pakistan matches it is AB Devilliers, Hashim Amla etc.

a <-teamBattingScorecardOppnAllMatches(aus_ind_matches,main="India",opposition="Australia")
## Total= 8331
a
## Source: local data frame [44 x 5]
## 
##         batsman ballsPlayed fours sixes  runs
##          (fctr)       (int) (int) (int) (dbl)
## 1      MS Dhoni        1406    78    22  1156
## 2     RG Sharma        1015    73    24   918
## 3  SR Tendulkar        1157   103     6   910
## 4       V Kohli         961    87     6   902
## 5     G Gambhir         677    44     2   536
## 6  Yuvraj Singh         664    52    11   524
## 7      SK Raina         536    43    11   509
## 8      S Dhawan         470    55     6   471
## 9      V Sehwag         305    42     4   289
## 10   RV Uthappa         295    29     7   283
## ..          ...         ...   ...   ...   ...
teamBattingScorecardOppnAllMatches(aus_ind_matches,"Australia","India")
## Total= 9995
## Source: local data frame [47 x 5]
## 
##       batsman ballsPlayed fours sixes  runs
##        (fctr)       (int) (int) (int) (dbl)
## 1  RT Ponting        1107    86     8   876
## 2  MEK Hussey         816    56     5   753
## 3   GJ Bailey         578    51    13   614
## 4   SR Watson         653    81    10   609
## 5   MJ Clarke         786    45     5   607
## 6   ML Hayden         660    72     8   573
## 7   A Symonds         543    43    15   536
## 8    AJ Finch         617    52     9   525
## 9   SPD Smith         431    44     7   467
## 10  DA Warner         385    40     6   391
## ..        ...         ...   ...   ...   ...
teamBattingScorecardOppnAllMatches(pak_sa_matches,"South Africa","Pakistan")
## Total= 6657
## Source: local data frame [36 x 5]
## 
##           batsman ballsPlayed fours sixes  runs
##            (fctr)       (int) (int) (int) (dbl)
## 1  AB de Villiers        1533   128    23  1423
## 2         HM Amla         864    88     3   815
## 3        GC Smith         726    68     3   597
## 4       JH Kallis         710    40     8   543
## 5       JP Duminy         620    35     3   481
## 6       CA Ingram         388    32     1   305
## 7    F du Plessis         363    30     4   278
## 8       Q de Kock         336    28     2   270
## 9       DA Miller         329    20     2   250
## 10       HH Gibbs         252    33     2   228
## ..            ...         ...   ...   ...   ...
teamBattingScorecardOppnAllMatches(sl_wi_matches,"West Indies","Sri Lanka")
## Total= 1800
## Source: local data frame [36 x 5]
## 
##          batsman ballsPlayed fours sixes  runs
##           (fctr)       (int) (int) (int) (dbl)
## 1       DM Bravo         353    20     6   265
## 2      RR Sarwan         315    11     3   205
## 3     MN Samuels         209    19     5   188
## 4       CH Gayle         198    18     8   176
## 5  S Chanderpaul         181     6     7   152
## 6      AB Barath         162     9     2   125
## 7       DJ Bravo         139     7     2   102
## 8       CS Baugh         102     5    NA    78
## 9    LMP Simmons          78     5     4    67
## 10     JO Holder          33     5     3    55
## ..           ...         ...   ...   ...   ...
teamBattingScorecardOppnAllMatches(eng_nz_matches,"England","New Zealand")
## Total= 6472
## Source: local data frame [47 x 5]
## 
##           batsman ballsPlayed fours sixes  runs
##            (fctr)       (int) (int) (int) (dbl)
## 1         IR Bell         871    74     7   654
## 2         JE Root         651    54     5   612
## 3  PD Collingwood         619    34    15   514
## 4      EJG Morgan         445    35    22   479
## 5         AN Cook         616    49     3   464
## 6       IJL Trott         421    26     1   362
## 7    KP Pietersen         481    30     6   358
## 8      JC Buttler         199    28    11   287
## 9         OA Shah         323    17     6   274
## 10      RS Bopara         350    21    NA   222
## ..            ...         ...   ...   ...   ...
teamBatsmenPartnershiOppnAllMatches(sco_can_matches,"Scotland","Canada")
## Source: local data frame [20 x 2]
## 
##          batsman totalRuns
##           (fctr)     (dbl)
## 1     CS MacLeod       177
## 2      MW Machan        68
## 3      CJO Smith        43
## 4    FRJ Coleman        40
## 5      RR Watson        14
## 6     JH Stander        12
## 7       MA Leask        12
## 8     RML Taylor        10
## 9     KJ Coetzer         8
## 10   GM Hamilton         7
## 11        RM Haq         7
## 12    PL Mommsen         6
## 13     CM Wright         5
## 14        JD Nel         5
## 15      MH Cross         4
## 16     SM Sharif         4
## 17     JAR Blain         2
## 18  NFI McCallum         1
## 19 RD Berrington         1
## 20     NS Poonia         0

9. Team performances of bowlers (all matches with opposition)

Like the function above the following tables provide the top bowlers of the countries in the matches against the oppoition. In India-Australia matches Ishant Sharma leads, in Pakistan-South Africa matches Shahid Afridi tops and so on.

teamBowlingPerfOppnAllMatches(aus_ind_matches,"India","Australia")
## Source: local data frame [36 x 5]
## 
##             bowler overs maidens  runs wickets
##             (fctr) (int)   (int) (dbl)   (dbl)
## 1         I Sharma    44       1   739      20
## 2  Harbhajan Singh    40       0   926      15
## 3        RA Jadeja    39       0   867      14
## 4        IK Pathan    42       1   702      11
## 5         UT Yadav    37       2   606      10
## 6          P Kumar    27       0   501      10
## 7           Z Khan    33       1   500      10
## 8      S Sreesanth    34       0   454      10
## 9         R Ashwin    43       0   684       9
## 10   R Vinay Kumar    31       1   380       9
## ..             ...   ...     ...   ...     ...
teamBowlingPerfOppnAllMatches(pak_sa_matches,main="Pakistan",opposition="South Africa")
## Source: local data frame [24 x 5]
## 
##             bowler overs maidens  runs wickets
##             (fctr) (int)   (int) (dbl)   (dbl)
## 1    Shahid Afridi    38       0  1053      17
## 2      Saeed Ajmal    39       0   658      14
## 3  Mohammad Hafeez    38       0   774      13
## 4   Mohammad Irfan    29       0   467      13
## 5   Iftikhar Anjum    29       1   257      12
## 6       Wahab Riaz    31       0   534      11
## 7      Junaid Khan    32       0   429      10
## 8    Sohail Tanvir    26       1   409       9
## 9    Shoaib Akhtar    22       1   313       9
## 10        Umar Gul    25       2   365       7
## ..             ...   ...     ...   ...     ...
teamBowlingPerfOppnAllMatches(eng_nz_matches,"New Zealand","England")
## Source: local data frame [33 x 5]
## 
##            bowler overs maidens  runs wickets
##            (fctr) (int)   (int) (dbl)   (dbl)
## 1      TG Southee    40       0   684      19
## 2        KD Mills    36       1   742      17
## 3      DL Vettori    35       0   561      16
## 4  MJ McClenaghan    34       0   515      14
## 5         SE Bond    17       1   205      11
## 6      GD Elliott    20       0   194      10
## 7    JEC Franklin    24       0   418       7
## 8   KS Williamson    21       1   225       7
## 9        TA Boult    18       2   195       7
## 10    NL McCullum    30       0   425       6
## ..            ...   ...     ...   ...     ...
teamBowlingPerfOppnAllMatches(sl_wi_matches,"Sri Lanka","West Indies")
## Source: local data frame [24 x 5]
## 
##             bowler overs maidens  runs wickets
##             (fctr) (int)   (int) (dbl)   (dbl)
## 1       SL Malinga    28       1   280      11
## 2       BAW Mendis    15       0   267       8
## 3  KMDN Kulasekara    13       1   185       7
## 4       AD Mathews    14       0   191       6
## 5   M Muralitharan    20       1   157       6
## 6      MF Maharoof     9       2    14       6
## 7       WPUJC Vaas     7       2    82       5
## 8       RAS Lakmal     7       0    55       4
## 9    ST Jayasuriya     1       0    38       4
## 10    HMRKB Herath    10       1   124       3
## ..             ...   ...     ...   ...     ...
teamBowlingPerfOppnAllMatches(ken_ber_matches,"Kenya","Bermuda")
## Source: local data frame [9 x 5]
## 
##        bowler overs maidens  runs wickets
##        (fctr) (int)   (int) (dbl)   (dbl)
## 1  JK Kamande    16       0   122       5
## 2  HA Varaiya    13       1    64       5
## 3   AS Luseno     6       0    32       4
## 4  PJ Ongondo     7       0    39       3
## 5    TM Odoyo     7       0    36       3
## 6  LN Onyango     7       0    37       2
## 7   SO Tikolo    18       0    81       1
## 8 NN Odhiambo    14       1    76       1
## 9    CO Obuya     4       0    20       0

10. Team bowler’s wickets (all matches with opposition)

This provided a graphical plot of the tables above

teamBowlersWicketsOppnAllMatches(aus_ind_matches,"India","Australia")

bowlerWicketsOppn-1

teamBowlersWicketsOppnAllMatches(aus_ind_matches,"Australia","India")

bowlerWicketsOppn-2

teamBowlersWicketsOppnAllMatches(pak_sa_matches,"South Africa","Pakistan",top=10)

bowlerWicketsOppn-3

m <-teamBowlersWicketsOppnAllMatches(eng_nz_matches,"England","Zealand",plot=FALSE)
m
## Source: local data frame [20 x 2]
## 
##            bowler wickets
##            (fctr)   (int)
## 1     JM Anderson      20
## 2       SCJ Broad      13
## 3         ST Finn      12
## 4  PD Collingwood      11
## 5        GP Swann      10
## 6   RJ Sidebottom       8
## 7       CR Woakes       8
## 8      A Flintoff       7
## 9     LE Plunkett       6
## 10      AU Rashid       6
## 11      BA Stokes       6
## 12     MS Panesar       5
## 13      LJ Wright       4
## 14     TT Bresnan       4
## 15      DJ Willey       4
## 16    JC Tredwell       3
## 17    CT Tremlett       2
## 18      RS Bopara       2
## 19      CJ Jordan       2
## 20        J Lewis       1
teamBowlersWicketsOppnAllMatches(ban_ire_matches,"Bangladesh","Ireland",top=7)

bowlerWicketsOppn-4

11. Team bowler vs batsmen (all matches with opposition)

These plots show how the bowlers fared against the batsmen. It shows which of the opposing teams batsmen were able to score the most runs

teamBowlersVsBatsmenOppnAllMatches(aus_ind_matches,'India',"Australia",top=5)

bowlerVsBatsmen-1

teamBowlersVsBatsmenOppnAllMatches(pak_sa_matches,"Pakistan","South Africa",top=3)

bowlerVsBatsmen-2

teamBowlersVsBatsmenOppnAllMatches(eng_nz_matches,"England","New Zealand")

bowlerVsBatsmen-3

teamBowlersVsBatsmenOppnAllMatches(eng_nz_matches,"New Zealand","England")

bowlerVsBatsmen-4

12. Team bowler’s wicket kind (caught,bowled,etc) (all matches with opposition)

The charts below show the wicket kind taken by the bowler (caught, bowled, lbw etc)

teamBowlersWicketKindOppnAllMatches(aus_ind_matches,"India","Australia",plot=TRUE)

bowlerWickets-1

m <- teamBowlersWicketKindOppnAllMatches(aus_ind_matches,"Australia","India",plot=FALSE)
m[1:30,]
##        bowler        wicketKind wicketPlayerOut runs
## 1  GD McGrath            caught    SR Tendulkar   69
## 2   SR Watson            caught        D Mongia  532
## 3  MG Johnson               lbw        V Sehwag 1020
## 4       B Lee            caught        R Dravid  671
## 5       B Lee            bowled          M Kaif  671
## 6  NW Bracken            caught        SK Raina  429
## 7  GD McGrath            caught       IK Pathan   69
## 8  NW Bracken               lbw        MS Dhoni  429
## 9  MG Johnson               lbw    SR Tendulkar 1020
## 10 MG Johnson            bowled       G Gambhir 1020
## 11   SR Clark            caught    SR Tendulkar  254
## 12   JR Hopes            caught    Yuvraj Singh  346
## 13   SR Clark               lbw      RV Uthappa  254
## 14    GB Hogg            caught        R Dravid  427
## 15  MJ Clarke           run out       IK Pathan  212
## 16  MJ Clarke           stumped Harbhajan Singh  212
## 17  MJ Clarke            bowled        RR Powar  212
## 18    GB Hogg            caught          Z Khan  427
## 19    GB Hogg            caught        MS Dhoni  427
## 20      B Lee               lbw       G Gambhir  671
## 21 MG Johnson               lbw      RV Uthappa 1020
## 22      B Lee            caught        R Dravid  671
## 23    GB Hogg            bowled    SR Tendulkar  427
## 24      B Lee            caught        MS Dhoni  671
## 25   JR Hopes            caught       RG Sharma  346
## 26    GB Hogg               lbw       IK Pathan  427
## 27 MG Johnson            bowled    Yuvraj Singh 1020
## 28    GB Hogg caught and bowled          Z Khan  427
## 29   SR Clark            bowled     S Sreesanth  254
## 30   JR Hopes            caught      SC Ganguly  346
teamBowlersWicketKindOppnAllMatches(sl_wi_matches,"Sri Lanka",'West Indies',plot=TRUE)

bowlerWickets-2

13. Team bowler’s wicket taken and runs conceded (all matches with opposition)

teamBowlersWicketRunsOppnAllMatches(aus_ind_matches,"India","Australia")

wicketRuns-1

m <-teamBowlersWicketRunsOppnAllMatches(pak_sa_matches,"Pakistan","South Africa",plot=FALSE)
m[1:30,]
## Source: local data frame [30 x 5]
## 
##             bowler overs maidens  runs wickets
##             (fctr) (int)   (int) (dbl)   (dbl)
## 1         Umar Gul    25       2   365       7
## 2   Iftikhar Anjum    29       1   257      12
## 3     Yasir Arafat     5       0    33       1
## 4     Abdul Razzaq    16       0   290       4
## 5  Mohammad Hafeez    38       0   774      13
## 6    Shahid Afridi    38       0  1053      17
## 7     Shoaib Malik    18       0   219       4
## 8    Sohail Tanvir    26       1   409       9
## 9     Abdur Rehman    25       0   301       4
## 10   Mohammad Asif    10       1   204       2
## ..             ...   ...     ...   ...     ...

14. Plot of wins vs losses between teams.

setwd("C:/software/cricket-package/york-test/yorkrData/ODI/ODI-matches")
plotWinLossBetweenTeams("India","Sri Lanka")

winsLosses-1

plotWinLossBetweenTeams('Pakistan',"South Africa",".")

winsLosses-2

plotWinLossBetweenTeams('England',"New Zealand",".")

winsLosses-3

plotWinLossBetweenTeams("Australia","West Indies",".")

winsLosses-4

plotWinLossBetweenTeams('Bangladesh',"Zimbabwe",".")

winsLosses-5

plotWinLossBetweenTeams('Scotland',"Ireland",".")

winsLosses-6

Conclusion

This post included all functions for all matches between any 2 opposing countries. As before the data frames are already available. You can load the data and begin to use them. If more insights from the dataframe are possible do go ahead. But please do attribute the source to Cricheet (http://cricsheet.org), my package yorkr and my blog. Do give the functions a spin for yourself.

There are 2 more posts required for the introduction of MY yorkr package.So, Hasta la vista, baby! I’ll be back!

Also see

You may also like

  1. Introducing cricketr! : An R package to analyze performances of cricketers
  2. Cricket analytics with cricketr
  3. cricketr adapts to the Twenty20 International!
  4. The making of Total Control Android game
  5. De-blurring revisited with Wiener filter using OpenCV
  6. Rock N’ Roll with Bluemix, Cloudant & NodeExpress

Cricket analytics with cricketr!!!


cricket

My ebook “Cricket analytics with cricketr’  has been published in Leanpub.  You can now download the book (hot off the press!)  for all formats to your favorite device (mobile, iPad, tablet, Kindle)  from the Leanpub  “Cricket analytics with cricketr”. The book has been published in the following formats namely

  • PDF (for your computer)
  • EPUB (for iPad or tablets. Save the file cricketr.epub to Google Drive/Dropbox and choose “Open in” iBooks for iPad)
  • MOBI (for Kindle. For this format, I suggest that you download & install SendToKindle for PC/Mac. You can then right click the downloaded cricketr.mobi and choose SendToKindle. You will need to login to your Kindle account)

From Leanpub
UntitledLeanpub uses a variable pricing model. I have priced the book attractively (I think!)  at $2.50 with a minimum price of $0.00 (FREE!!! limited time offer!).  The link is “Cricket analytics with cricketr

This format works with all type Kindle, Android tablet, iPad.

From Amazon
UntitledYou can also download for Kindle. The price is $2.50 (Rs 169/-). Cricket analytics with cricketr.

Do download the book and hope you have many happy hours reading it.

I am including my preface in the book below

Preface
Cricket has been the “national passion” of India for decades. As a boy I was also held in thrall by a strong cricketing passion like many. Cricket is a truly fascinating game! I would catch the sporting action with my friends as we crowded around a transistor that brought us live, breathless radio commentary. We also spent many hours glued to live cricket action on the early black and white TVs. This used to be an experience of sorts, as every now and then a part of the body of the players, would detach itself and stretch to the sides. But it was enjoyable all the same.

Nowadays broadcast technology has improved so much and we get detailed visual analysis of the how each bowler varies the swing and length of the delivery. We are also able to see the strokes of batsman in slow motion.   Similarly computing technology has also advanced by leaps and bounds and we can analyze players in great detail with a few lines of code in languages like R, Python etc.

In 2015, I completed Machine Learning from Stanford at Coursera.  I was looking around for data to play around with, when it suddenly struck me that I could do some regression analysis of batting records.  In the subsequent months, I took the Data Science Specialization from John Hopkins University, which triggered more ideas in me. One thing led to another and I managed to put together an R package called ‘cricketr’.  I developed this package over 7 months adding and refining functions. Finally, I managed to submit the package to CRAN.  During the development of the package for different formats of the game I wrote a series of posts in my blog.

This book is a collection of those cricket related posts.  There are 6 posts based on my R package cricketr. I have also included 2 earlier posts based on R which I wrote before I created my R package. Finally, I also include another 2 cricket posts based on Machine Learning in which I used the language Octave.

My cricketr’ package is a first, for cricket analytics, howzzat!  and I am certain that it won’t be the last. Cricket is a wonderful pitch for statisticians, data scientists and machine learning experts. So you can expect some cool packages in the years to come.

I had a great time developing the package. I hope you have a wonderful time reading this book. Do remember to download from “Cricket analytics with cricketr

Feel free to get in touch with me anytime through email included below

Tinniam V Ganesh
tvganesh.85@gmail.com
January 28, 2016

cricketr adapts to the Twenty20 International!


Introduction

This should be last in the series of posts based on my R package cricketr. That is, unless some bright idea comes trotting along and light bulbs go on around my head.

In this post cricketr adapts to the Twenty20 International format. Now cricketr can handle stats from all 3 formats of the game namely Test matches, ODIs and Twenty20 International from ESPN Cricinfo. You should be able to install the package from GitHub and use the many of the functions available in the package.

Please be mindful of the ESPN Cricinfo Terms of Use

You can also read this post at Rpubs as twenty20-cricketr. Download this report as a PDF file from twenty20-cricketr.pdf

Do check out my interactive Shiny app implementation using the cricketr package – Sixer – R package cricketr’s new Shiny avatar

Check out my 2 books on cricket, a) Cricket analytics with cricketr b) Beaten by sheer pace – Cricket analytics with yorkr, now available in both paperback & kindle versions on Amazon!!! Pick up your copies today!

Note: If you would like to do a similar analysis for a different set of batsman and bowlers, you can clone/download my skeleton cricketr template from Github (which is the R Markdown file I have used for the analysis below). You will only need to make appropriate changes for the players you are interested in. Just a familiarity with R and R Markdown only is needed.

I have chosen the Top 4 batsmen and top 4 bowlers based on ICC rankings and/or number of matches played.

Batsmen

  1. Virat Kohli (Ind)
  2. Faf du Plessis (SA)
  3. A J Finch (Aus)
  4. Brendon McCullum (Aus)

Bowlers

  1. Samuel Badree (WI)
  2. Sunil Narine (WI)
  3. Ravichander Ashwin (Ind)
  4. Ajantha Mendis (SL)

I have explained the plots and added my own observations. Please feel free to draw your conclusions!

The data for a particular player can be obtained with the getPlayerData() function. To do you will need to go to ESPN CricInfo Player and type in the name of the player for e.g Virat Kohli, Sunil Narine etc. This will bring up a page which have the profile number for the player e.g. for Virat Kohli this would be http://www.espncricinfo.com/india/content/player/253802.html.

The package can be installed directly from CRAN

if (!require("cricketr")){ 
    install.packages("cricketr",lib = "c:/test") 
} 
library(cricketr)

or from Github

library(devtools)
install_github("tvganesh/cricketr")
library(cricketr)

The data for a particular player can be obtained with the getPlayerData() function. To do you will need to go to ESPN CricInfo Player and type in the name of the player for e.g Virat Kohli, Sunil Narine etc. This will bring up a page which have the profile number for the player e.g. for Virat Kohli this would be http://www.espncricinfo.com/india/content/player/253802.html. Hence, Kohlis profile is 253802. This can be used to get the data for Virat Kohli as shown below

kohli <- getPlayerDataTT(253802,dir="..",file="kohli.csv",type="batting")

The analysis is included below

Analyses of Batsmen

The following plots gives the analysis of the 4 ODI batsmen

  1. Virat Kohli (Ind) – Innings-26, Runs-972, Average-46.28,Strike Rate-131.70
  2. Faf du Plessis (SA) – Innings-24, Runs-805, Average-42.36,Strike Rate-135.75
  3. A J Finch (Aus) – Innings-22, Runs-756, Average-39.78,Strike Rate-152.41
  4. Brendon McCullum (NZ) – Innings-70, Runs-2140, Average-35.66,Strike Rate-136.21

Plot of 4s, 6s and the scoring rate in ODIs

The 3 charts below give the number of

  1. 4s vs Runs scored
  2. 6s vs Runs scored
  3. Balls faced vs Runs scored A regression line is fitted in each of these plots for each of the ODI batsmen

A. Virat Kohli
– The 1st plot shows that Kohli approximately hits about 5 4’s on his way to the 50s
– The 2nd box plot of no of 6s and runs shows the range of runs when Kohli scored 1,2 or 4 6s. The dark line in the box shows the average runs when he scored those number of 6s. So when he scored 1 6 the average runs he scored was 45
– The 3rd plot shows the number of runs scored against the balls faced. It can be seen when Kohli faced 50 balls he had scored around ~ 70 runs

par(mfrow=c(1,3))
par(mar=c(4,4,2,2))
batsman4s("./kohli.csv","Kohli")
batsman6s("./kohli.csv","Kohli")
batsmanScoringRateODTT("./kohli.csv","Kohli")

kohli-4s6sSR-1

dev.off()
## null device 
##           1

B. Faf du Plessis

par(mfrow=c(1,3))
par(mar=c(4,4,2,2))
batsman4s("./plessis.csv","Du Plessis")
batsman6s("./plessis.csv","Du Plessis")
batsmanScoringRateODTT("./plessis.csv","Du Plessss")

plessis-4s6SR-1

dev.off()
## null device 
##           1

C. A J Finch

par(mfrow=c(1,3))
par(mar=c(4,4,2,2))
batsman4s("./finch.csv","A J Finch")
batsman6s("./finch.csv","A J Finch")
batsmanScoringRateODTT("./finch.csv","A J Finch")

finch-4s6sSR-1

dev.off()
## null device 
##           1

D. Brendon McCullum

par(mfrow=c(1,3))
par(mar=c(4,4,2,2))
batsman4s("./mccullum.csv","McCullum")
batsman6s("./mccullum.csv","McCullum")
batsmanScoringRateODTT("./mccullum.csv","McCullum")

mccullum-4s6sout-1

dev.off()
## null device 
##           1

Relative Mean Strike Rate

This plot shows the Mean Strike Rate of the batsman in each run range. It can be seen the A J Finch has the best strike rate followed by B McCullum.

par(mar=c(4,4,2,2))
frames <- list("./kohli.csv","./plessis.csv","finch.csv","mccullum.csv")
names <- list("Kohli","Du Plessis","Finch","McCullum")
relativeBatsmanSRODTT(frames,names)

plot-1-1

Relative Runs Frequency Percentage

The plot below provides the average runs scored in each run range 0-5,5-10,10-15 etc. Clearly Kohli has the most runs scored in most of the runs ranges. . This is also evident in the fact that Kohli has the highest average. He is followed by McCullum

frames <- list("./kohli.csv","./plessis.csv","finch.csv","mccullum.csv")
names <- list("Kohli","Du Plessis","Finch","McCullum")
relativeRunsFreqPerfODTT(frames,names)

plot-2-1

Percent 4’s,6’s in total runs scored

The plot below shows the percentage of runs scored by way of 4s and 6s for each batsman. Du Plessis has the highest percentage of 4s, McCullum has the highest 6s. Finch has the highest percentage of 4s & 6s – 25.37 + 15.64= 41.01%

rames <- list("./kohli.csv","./plessis.csv","finch.csv","mccullum.csv")
names <- list("Kohli","Du Plessis","Finch","McCullum")
runs4s6s <-batsman4s6s(frames,names)

plot-46s-1

print(runs4s6s)
##                Kohli Du Plessis Finch McCullum
## Runs(1s,2s,3s) 64.29      64.55 58.99    61.45
## 4s             27.78      24.38 25.37    22.87
## 6s              7.94      11.07 15.64    15.69

3D plot of Runs vs Balls Faced and Minutes at Crease

The plot is a scatter plot of Runs vs Balls faced and Minutes at Crease. A prediction plane is then fitted based on the Balls Faced and Minutes at Crease to give the runs scored

par(mfrow=c(1,2))
par(mar=c(4,4,2,2))
battingPerf3d("./kohli.csv","Kohli")
battingPerf3d("./plessis.csv","Du Plessis")

plot-3-1

dev.off()
## null device 
##           1
par(mfrow=c(1,2))
par(mar=c(4,4,2,2))
battingPerf3d("./finch.csv","A J Finch")
battingPerf3d("./mccullum.csv","McCullum")

plot-4-1

dev.off()
## null device 
##           1

Predicting Runs given Balls Faced and Minutes at Crease

A hypothetical Balls faced and Minutes at Crease is used to predict the runs scored by each batsman based on the computed prediction plane

BF <- seq( 5, 70,length=10)
Mins <- seq(5,70,length=10)
newDF <- data.frame(BF,Mins)

kohli <- batsmanRunsPredict("./kohli.csv","Kohli",newdataframe=newDF)
plessis <- batsmanRunsPredict("./plessis.csv","Du Plessis",newdataframe=newDF)
finch <- batsmanRunsPredict("./finch.csv","A J Finch",newdataframe=newDF)
mccullum <- batsmanRunsPredict("./mccullum.csv","McCullum",newdataframe=newDF)

The predicted runs is displayed. As can be seen Finch has the best overall strike rate followed by McCullum.

batsmen <-cbind(round(kohli$Runs),round(plessis$Runs),round(finch$Runs),round(mccullum$Runs))
colnames(batsmen) <- c("Kohli","Du Plessis","Finch","McCullum")
newDF <- data.frame(round(newDF$BF),round(newDF$Mins))
colnames(newDF) <- c("BallsFaced","MinsAtCrease")
predictedRuns <- cbind(newDF,batsmen)
predictedRuns
##    BallsFaced MinsAtCrease Kohli Du Plessis Finch McCullum
## 1           5            5     2          1     5        3
## 2          12           12    12         10    22       16
## 3          19           19    22         19    40       28
## 4          27           27    31         28    57       41
## 5          34           34    41         37    74       54
## 6          41           41    51         47    91       66
## 7          48           48    60         56   108       79
## 8          56           56    70         65   125       91
## 9          63           63    79         74   142      104
## 10         70           70    89         84   159      117

Highest runs likelihood

The plots below the runs likelihood of batsman. This uses K-Means Kohli has the highest likelihood of scoring runs 34.2% likely to score 66 runs. Du Plessis has 25% likelihood to score 53 runs, A. Virat Kohli

batsmanRunsLikelihood("./kohli.csv","Kohli")

kohli-lh-1

## Summary of  Kohli 's runs scoring likelihood
## **************************************************
## 
## There is a 23.08 % likelihood that Kohli  will make  10 Runs in  10 balls over 13  Minutes 
## There is a 42.31 % likelihood that Kohli  will make  29 Runs in  23 balls over  30  Minutes 
## There is a 34.62 % likelihood that Kohli  will make  66 Runs in  47 balls over 63  Minutes

B. Faf Du Plessis

batsmanRunsLikelihood("./plessis.csv","Du Plessis")

plessis-l-1

## Summary of  Du Plessis 's runs scoring likelihood
## **************************************************
## 
## There is a 62.5 % likelihood that Du Plessis  will make  14 Runs in  11 balls over 19  Minutes 
## There is a 25 % likelihood that Du Plessis  will make  53 Runs in  40 balls over  50  Minutes 
## There is a 12.5 % likelihood that Du Plessis  will make  94 Runs in  61 balls over 90  Minutes

C. A J Finch

batsmanRunsLikelihood("./finch.csv","A J Finch")

finch-lh,cache-TRUE-1

## Summary of  A J Finch 's runs scoring likelihood
## **************************************************
## 
## There is a 20 % likelihood that A J Finch  will make  95 Runs in  54 balls over 70  Minutes 
## There is a 25 % likelihood that A J Finch  will make  42 Runs in  27 balls over  35  Minutes 
## There is a 55 % likelihood that A J Finch  will make  8 Runs in  8 balls over 12  Minutes

D. Brendon McCullum

batsmanRunsLikelihood("./mccullum.csv","McCullum")

mccullum-1

## Summary of  McCullum 's runs scoring likelihood
## **************************************************
## 
## There is a 50.72 % likelihood that McCullum  will make  11 Runs in  10 balls over 13  Minutes 
## There is a 28.99 % likelihood that McCullum  will make  36 Runs in  27 balls over  37  Minutes 
## There is a 20.29 % likelihood that McCullum  will make  74 Runs in  48 balls over 70  Minutes

Moving Average of runs over career

The moving average for the 4 batsmen indicate the following. It must be noted that there is not sufficient data yet on Twenty20 Internationals. Kpohli, Du Plessis and Finch average only 26 innings while McCullum has close to 70. So the moving average while an indication will regress towards the mean over time.

  1. The moving average of Kohli and Du Plessis is on the way up.
  2. McCullum has a consistent performance while Finch had a brief burst in 2013-2014
par(mfrow=c(2,2))
par(mar=c(4,4,2,2))
batsmanMovingAverage("./kohli.csv","Kohli")
batsmanMovingAverage("./plessis.csv","Du Plessis")
batsmanMovingAverage("./finch.csv","A J Finch")
batsmanMovingAverage("./mccullum.csv","McCullum")

sdgm-ma-1

dev.off()
## null device 
##           1

Analysis of bowlers

  1. Samuel Badree (WI) – Innings-22, Runs -464, Wickets – 31, Econ Rate : 5.39
  2. Sunil Narine (WI)- Innings-31,Runs-666, Wickets – 38 , Econ Rate : 5.70
  3. Ravichander Ashwin (Ind)- Innings-26, Runs- 732, Wickets – 25, Econ Rate : 7.32
  4. Ajantha Mendis (SL)- Innings-39, Runs – 952,Wickets – 66, Econ Rate : 6.45

The plot shows the frequency with which the bowlers have taken 1,2,3 etc wickets. The most wickets taken is by Ajantha Mendis (6 wickets)

Wicket Frequency percentage

This plot gives the percentage of wickets for each wickets (1,2,3…etc)

par(mfrow=c(1,4))
par(mar=c(4,4,2,2))
bowlerWktsFreqPercent("./badree.csv","Badree")
bowlerWktsFreqPercent("./mendis.csv","Mendis")
bowlerWktsFreqPercent("./narine.csv","Narine")
bowlerWktsFreqPercent("./ashwin.csv","Ashwin")

relBowlFP-1

dev.off()
## null device 
##           1

Wickets Runs plot

The plot below gives a boxplot of the runs ranges for each of the wickets taken by the bowlers. The ends of the box indicate the 25% and 75% percentile of runs scored for the wickets taken and the dark balck line is the average runs conceded.

par(mfrow=c(1,4))
par(mar=c(4,4,2,2))
bowlerWktsRunsPlot("./badree.csv","Badree")
bowlerWktsRunsPlot("./mendis.csv","Mendis")
bowlerWktsRunsPlot("./narine.csv","Narine")
bowlerWktsRunsPlot("./ashwin.csv","Ashwin")

wktsrun-1

dev.off()
## null device 
##           1

This plot below shows the average number of deliveries needed by the bowler to take the wickets (1,2,3 etc)

par(mfrow=c(1,2))
par(mar=c(4,4,2,2))
bowlerWktRateTT("./badree.csv","Badree")
bowlerWktRateTT("./mendis.csv","Mendis")

wktsrate1-1

dev.off()
## null device 
##           1
par(mfrow=c(1,2))
par(mar=c(4,4,2,2))
bowlerWktRateTT("./narine.csv","Narine")
bowlerWktRateTT("./ashwin.csv","Ashwin")

wktsrate2-1

dev.off()
## null device 
##           1

Relative bowling performance

The plot below shows that Narine has the most wickets in the 2 -4 range followed by Mendis

frames <- list("./badree.csv","./mendis.csv","narine.csv","ashwin.csv")
names <- list("Badree","Mendis","Narine","Ashwin")
relativeBowlingPerf(frames,names)

relBowlPerf-1

Relative Economy Rate against wickets taken

The economy rate can be deduced as follows from the plot below. Narine has a good economy rate around 1 & 4 wickets, Ashwin around 2 wickets and Badree around 3. wickets

frames <- list("./badree.csv","./mendis.csv","narine.csv","ashwin.csv")
names <- list("Badree","Mendis","Narine","Ashwin")
relativeBowlingERODTT(frames,names)

relBowlER-1

Relative Wicket Rate

The relative wicket rate plots the mean number of deliveries needed to take the wickets namely (1,2,3,4). For e.g. Narine needed an average of 22 deliveries to take 1 wicket and 22.5,23.2, 24 deliveries to take 2,3 & 4 wickets respectively

frames <- list("./badree.csv","./mendis.csv","narine.csv","ashwin.csv")
names <- list("Badree","Mendis","Narine","Ashwin")
relativeWktRateTT(frames,names)

relBowlWktRate-1

Moving average of wickets over career

par(mfrow=c(2,2))
par(mar=c(4,4,2,2))
bowlerMovingAverage("./badree.csv","Badree")
bowlerMovingAverage("./mendis.csv","Mendis")
bowlerMovingAverage("./narine.csv","Narine")
bowlerMovingAverage("./ashwin.csv","Ashwin")
## null device 
##           1

jsba-bowlma-1

Key findings

Here are some key conclusions

Twenty 20 batsmen

  1. Kohli has the a very consistent performance scoring high runs in the different run ranges. Kohli also has a 34.2% likelihood to score 6 runs. He is followed by McCullum for consisten performance
  2. Finch has a best strike rate followed by McCullum.
  3. Du Plessis has the highest percentage of 4s and McCullum has the percentage of 6s. Finch is superior in the percentage of runs scored in 4s and 6s
  4. For a hypothetical balls faced and minutes at crease, Finch does best followed by McCullum
  5. Kohli’s & Du Plessis Twenty20 career is on a upswing. Can they maintain the momentum. McCullum is consistent

Twenty20 bowlers

  1. Narine has the highest wickets percentage for different wickets taken followed by Mendis
  2. Mendis has taken 1,2,3,4,6 wickets in 24 deliveries
  3. Narine has the lowest economy rate for 1 & 4 wickets, Ashwin for 2 wickets and Badree for 3 wickets. Mendis is comparatively expensive
  4. Narine needed the least deliveries to get 1 (22.5) & 2 (23.2) wickets, Mendis needed 20.5 deliveries and Ashwin 19 deliveries for 4 wickets

Key takeaways 1. If all the above batsment and bowlers were in the same team we expect

  1. Finch would be most useful when the run rate has to be greatly accelerated followed by McCullum
  2. If the need is to consolidate, then Kohli is the best man for the job followed by McCullum
  3. Overall McCullum is the best bet for Twenty20
  4. When it comes to bowling Narine wins hands down as he has the most wickets, a good economy rate and a very good attack rate. So Narine is great bet for providing a vital breakthrough.

Also see my other posts in R

  1. Introducing cricketr! : An R package to analyze performances of cricketers
  2. cricketr plays the ODIs!
  3. A peek into literacy in India: Statistical Learning with R
  4. A crime map of India in R – Crimes against women
  5. Analyzing cricket’s batting legends – Through the mirage with R
  6. Mirror, mirror . the best batsman of them all?

You may also like

  1. A closer look at “Robot Horse on a Trot” in Android
  2. What’s up Watson? Using IBM Watson’s QAAPI with Bluemix, NodeExpress – Part 1
  3. Bend it like Bluemix, MongoDB with autoscaling – Part 2
  4. Informed choices through Machine Learning : Analyzing Kohli, Tendulkar and Dravid
  5. TWS-4: Gossip protocol: Epidemics and rumors to the rescue
  6. Deblurring with OpenCV:Weiner filter reloaded
  7. Architecting a cloud based IP Multimedia System (IMS)