Scribd
Upload
4K views15 pages

Variable Selection

The document analyzes baseball player salary data using variable selection methods in R. It uses best subset selection, forward stepwise selection, and backward stepwise selection to identify the most predictive variables. For each method, it examines metrics like RSS, adjusted R-squared, Cp, and BIC to select the optimal number of variables. The analysis identifies variables like AtBat, Hits, Runs, RBI, and Walks as consistently predictive of salary across different selection techniques.

Uploaded by

api-285777244
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4K views15 pages

Variable Selection

The document analyzes baseball player salary data using variable selection methods in R. It uses best subset selection, forward stepwise selection, and backward stepwise selection to identify the most predictive variables. For each method, it examines metrics like RSS, adjusted R-squared, Cp, and BIC to select the optimal number of variables. The analysis identifies variables like AtBat, Hits, Runs, RBI, and Walks as consistently predictive of salary across different selection techniques.

Uploaded by

api-285777244
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 15

Variable Selection

YIK LUN, KEI


[email protected]
This paper is a lab from the book called An Introduction to Statistical Learning
with Applications in R. All R codes and comments below are belonged to the
book and authors.

Best Subset Selection


library(ISLR)
library(leaps)
Hitters<-na.omit(Hitters)
dim(Hitters)
## [1] 263

20

regfit.full=regsubsets (Salary~.,Hitters,nvmax=19)
reg.summary<-summary(regfit.full)
names(reg.summary)

## [1] "which"

"rsq"

"rss"

"adjr2"

"cp"

"bic"

"outmat" "obj"

RSS
plot(reg.summary$rss,xlab="Number of Variables",ylab="RSS",type="l")

3.6e+07
3.2e+07
2.8e+07

RSS

2.4e+07

10

15

Number of Variables

Adjusted R-squared
which.max(reg.summary$adjr2)

## [1] 11
plot(reg.summary$adjr2,xlab ="Number of Variables",ylab="Adjusted RSq",type="l")
points(11,reg.summary$adjr2[11],col="red",cex=2, pch=20)

0.50
0.45
0.40
0.35

Adjusted RSq

10

15

Number of Variables

Cp
which.min(reg.summary$cp)

## [1] 10
plot(reg.summary$cp,xlab ="Number of Variables",ylab="Cp",type="l")
points(10,reg.summary$cp[10],col ="red",cex=2, pch=20)

100
80
60
20

40

Cp

10

15

Number of Variables

BIC
which.min(reg.summary$bic)

## [1] 6
plot(reg.summary$bic,xlab="Number of Variables",ylab="BIC",type="l")
points (6,reg.summary$bic[6],col ="red",cex=2,pch=20)

90
110
150

130

BIC

10
Number of Variables

Selected Variables
plot(regfit.full,scale ="r2")

15

(Intercept)
AtBat
Hits
HmRun
Runs
RBI
Walks
Years
CAtBat
CHits
CHmRun
CRuns
CRBI
CWalks
LeagueN
DivisionW
PutOuts
Assists
Errors
NewLeagueN

r2
0.55
0.55
0.55
0.55
0.55
0.55
0.54
0.54
0.54
0.54
0.53
0.53
0.51
0.51
0.49
0.48
0.45
0.43
0.32

plot(regfit.full,scale ="adjr2")

(Intercept)
AtBat
Hits
HmRun
Runs
RBI
Walks
Years
CAtBat
CHits
CHmRun
CRuns
CRBI
CWalks
LeagueN
DivisionW
PutOuts
Assists
Errors
NewLeagueN

adjr2
0.52
0.52
0.52
0.52
0.52
0.52
0.52
0.52
0.51
0.51
0.51
0.51
0.5
0.5
0.48
0.47
0.45
0.42
0.32

plot(regfit.full,scale ="Cp")

(Intercept)
AtBat
Hits
HmRun
Runs
RBI
Walks
Years
CAtBat
CHits
CHmRun
CRuns
CRBI
CWalks
LeagueN
DivisionW
PutOuts
Assists
Errors
NewLeagueN

Cp
5
5.9
6.2
7.3
7.4
8.9
10
12
13
14
14
16
18
20
22
28
39
51
100

plot(regfit.full,scale ="bic")

(Intercept)
AtBat
Hits
HmRun
Runs
RBI
Walks
Years
CAtBat
CHits
CHmRun
CRuns
CRBI
CWalks
LeagueN
DivisionW
PutOuts
Assists
Errors
NewLeagueN

bic

150
150
150
150
140
140
140
140
140
130
130
130
120
120
110
110
100
96
91

coef(regfit.full,6) #by BIC


## (Intercept)
##
91.5117981
##
DivisionW
## -122.9515338

AtBat
-1.8685892
PutOuts
0.2643076

Hits
7.6043976

Walks
3.6976468

CRBI
0.6430169

Forward Stepwise Selection


regfit.fwd=regsubsets(Salary~.,data=Hitters,nvmax =19,method ="forward")
summary(regfit.fwd)

##
##
##
##
##
##
##
##
##
##

Subset selection object


Call: regsubsets.formula(Salary ~ ., data = Hitters, nvmax = 19, method = "forward")
19 Variables (and intercept)
Forced in Forced out
AtBat
FALSE
FALSE
Hits
FALSE
FALSE
HmRun
FALSE
FALSE
Runs
FALSE
FALSE
RBI
FALSE
FALSE
Walks
FALSE
FALSE
9

##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##

Years
FALSE
FALSE
CAtBat
FALSE
FALSE
CHits
FALSE
FALSE
CHmRun
FALSE
FALSE
CRuns
FALSE
FALSE
CRBI
FALSE
FALSE
CWalks
FALSE
FALSE
LeagueN
FALSE
FALSE
DivisionW
FALSE
FALSE
PutOuts
FALSE
FALSE
Assists
FALSE
FALSE
Errors
FALSE
FALSE
NewLeagueN
FALSE
FALSE
1 subsets of each size up to 19
Selection Algorithm: forward
AtBat Hits HmRun Runs RBI Walks Years
1 ( 1 ) " "
" " " "
" " " " " "
" "
2 ( 1 ) " "
"*" " "
" " " " " "
" "
3 ( 1 ) " "
"*" " "
" " " " " "
" "
4 ( 1 ) " "
"*" " "
" " " " " "
" "
5 ( 1 ) "*"
"*" " "
" " " " " "
" "
6 ( 1 ) "*"
"*" " "
" " " " "*"
" "
7 ( 1 ) "*"
"*" " "
" " " " "*"
" "
8 ( 1 ) "*"
"*" " "
" " " " "*"
" "
9 ( 1 ) "*"
"*" " "
" " " " "*"
" "
10 ( 1 ) "*"
"*" " "
" " " " "*"
" "
11 ( 1 ) "*"
"*" " "
" " " " "*"
" "
12 ( 1 ) "*"
"*" " "
"*" " " "*"
" "
13 ( 1 ) "*"
"*" " "
"*" " " "*"
" "
14 ( 1 ) "*"
"*" "*"
"*" " " "*"
" "
15 ( 1 ) "*"
"*" "*"
"*" " " "*"
" "
16 ( 1 ) "*"
"*" "*"
"*" "*" "*"
" "
17 ( 1 ) "*"
"*" "*"
"*" "*" "*"
" "
18 ( 1 ) "*"
"*" "*"
"*" "*" "*"
"*"
19 ( 1 ) "*"
"*" "*"
"*" "*" "*"
"*"
CRBI CWalks LeagueN DivisionW PutOuts
1 ( 1 ) "*" " "
" "
" "
" "
2 ( 1 ) "*" " "
" "
" "
" "
3 ( 1 ) "*" " "
" "
" "
"*"
4 ( 1 ) "*" " "
" "
"*"
"*"
5 ( 1 ) "*" " "
" "
"*"
"*"
6 ( 1 ) "*" " "
" "
"*"
"*"
7 ( 1 ) "*" "*"
" "
"*"
"*"
8 ( 1 ) "*" "*"
" "
"*"
"*"
9 ( 1 ) "*" "*"
" "
"*"
"*"
10 ( 1 ) "*" "*"
" "
"*"
"*"
11 ( 1 ) "*" "*"
"*"
"*"
"*"
12 ( 1 ) "*" "*"
"*"
"*"
"*"
13 ( 1 ) "*" "*"
"*"
"*"
"*"
14 ( 1 ) "*" "*"
"*"
"*"
"*"
15 ( 1 ) "*" "*"
"*"
"*"
"*"
16 ( 1 ) "*" "*"
"*"
"*"
"*"
17 ( 1 ) "*" "*"
"*"
"*"
"*"
18 ( 1 ) "*" "*"
"*"
"*"
"*"

10

CAtBat CHits CHmRun CRuns


" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
"*"
" "
"*"
"*"
"*"
" "
"*"
"*"
"*"
" "
"*"
"*"
"*"
" "
"*"
"*"
"*"
"*"
"*"
Assists Errors NewLeagueN
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
"*"
" "
" "
"*"
" "
" "
"*"
" "
" "
"*"
"*"
" "
"*"
"*"
" "
"*"
"*"
" "
"*"
"*"
" "
"*"
"*"
"*"
"*"
"*"
"*"

## 19

( 1 ) "*"

"*"

"*"

"*"

"*"

"*"

"*"

"*"

150
150
150
140
140
140
140
140
140
130
130
130
120
120
110
110
100
96
91
(Intercept)
AtBat
Hits
HmRun
Runs
RBI
Walks
Years
CAtBat
CHits
CHmRun
CRuns
CRBI
CWalks
LeagueN
DivisionW
PutOuts
Assists
Errors
NewLeagueN

bic

plot(regfit.fwd,scale="bic")

Backward Stepwise Selection


regfit.bwd=regsubsets(Salary~.,data=Hitters,nvmax =19,method ="backward")
summary(regfit.bwd)

##
##
##
##
##
##
##
##
##
##
##
##
##

Subset selection object


Call: regsubsets.formula(Salary ~ ., data = Hitters, nvmax = 19, method = "backward")
19 Variables (and intercept)
Forced in Forced out
AtBat
FALSE
FALSE
Hits
FALSE
FALSE
HmRun
FALSE
FALSE
Runs
FALSE
FALSE
RBI
FALSE
FALSE
Walks
FALSE
FALSE
Years
FALSE
FALSE
CAtBat
FALSE
FALSE
CHits
FALSE
FALSE
11

##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##
##

CHmRun
FALSE
FALSE
CRuns
FALSE
FALSE
CRBI
FALSE
FALSE
CWalks
FALSE
FALSE
LeagueN
FALSE
FALSE
DivisionW
FALSE
FALSE
PutOuts
FALSE
FALSE
Assists
FALSE
FALSE
Errors
FALSE
FALSE
NewLeagueN
FALSE
FALSE
1 subsets of each size up to 19
Selection Algorithm: backward
AtBat Hits HmRun Runs RBI Walks Years
1 ( 1 ) " "
" " " "
" " " " " "
" "
2 ( 1 ) " "
"*" " "
" " " " " "
" "
3 ( 1 ) " "
"*" " "
" " " " " "
" "
4 ( 1 ) "*"
"*" " "
" " " " " "
" "
5 ( 1 ) "*"
"*" " "
" " " " "*"
" "
6 ( 1 ) "*"
"*" " "
" " " " "*"
" "
7 ( 1 ) "*"
"*" " "
" " " " "*"
" "
8 ( 1 ) "*"
"*" " "
" " " " "*"
" "
9 ( 1 ) "*"
"*" " "
" " " " "*"
" "
10 ( 1 ) "*"
"*" " "
" " " " "*"
" "
11 ( 1 ) "*"
"*" " "
" " " " "*"
" "
12 ( 1 ) "*"
"*" " "
"*" " " "*"
" "
13 ( 1 ) "*"
"*" " "
"*" " " "*"
" "
14 ( 1 ) "*"
"*" "*"
"*" " " "*"
" "
15 ( 1 ) "*"
"*" "*"
"*" " " "*"
" "
16 ( 1 ) "*"
"*" "*"
"*" "*" "*"
" "
17 ( 1 ) "*"
"*" "*"
"*" "*" "*"
" "
18 ( 1 ) "*"
"*" "*"
"*" "*" "*"
"*"
19 ( 1 ) "*"
"*" "*"
"*" "*" "*"
"*"
CRBI CWalks LeagueN DivisionW PutOuts
1 ( 1 ) " " " "
" "
" "
" "
2 ( 1 ) " " " "
" "
" "
" "
3 ( 1 ) " " " "
" "
" "
"*"
4 ( 1 ) " " " "
" "
" "
"*"
5 ( 1 ) " " " "
" "
" "
"*"
6 ( 1 ) " " " "
" "
"*"
"*"
7 ( 1 ) " " "*"
" "
"*"
"*"
8 ( 1 ) "*" "*"
" "
"*"
"*"
9 ( 1 ) "*" "*"
" "
"*"
"*"
10 ( 1 ) "*" "*"
" "
"*"
"*"
11 ( 1 ) "*" "*"
"*"
"*"
"*"
12 ( 1 ) "*" "*"
"*"
"*"
"*"
13 ( 1 ) "*" "*"
"*"
"*"
"*"
14 ( 1 ) "*" "*"
"*"
"*"
"*"
15 ( 1 ) "*" "*"
"*"
"*"
"*"
16 ( 1 ) "*" "*"
"*"
"*"
"*"
17 ( 1 ) "*" "*"
"*"
"*"
"*"
18 ( 1 ) "*" "*"
"*"
"*"
"*"
19 ( 1 ) "*" "*"
"*"
"*"
"*"

12

CAtBat CHits CHmRun CRuns


" "
" "
" "
"*"
" "
" "
" "
"*"
" "
" "
" "
"*"
" "
" "
" "
"*"
" "
" "
" "
"*"
" "
" "
" "
"*"
" "
" "
" "
"*"
" "
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
" "
" "
"*"
"*"
"*"
" "
"*"
"*"
"*"
" "
"*"
"*"
"*"
" "
"*"
"*"
"*"
" "
"*"
"*"
"*"
"*"
"*"
Assists Errors NewLeagueN
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
" "
"*"
" "
" "
"*"
" "
" "
"*"
" "
" "
"*"
"*"
" "
"*"
"*"
" "
"*"
"*"
" "
"*"
"*"
" "
"*"
"*"
"*"
"*"
"*"
"*"
"*"
"*"
"*"

150
150
140
140
140
140
140
140
130
130
130
120
120
120
110
110
100
96
89
(Intercept)
AtBat
Hits
HmRun
Runs
RBI
Walks
Years
CAtBat
CHits
CHmRun
CRuns
CRBI
CWalks
LeagueN
DivisionW
PutOuts
Assists
Errors
NewLeagueN

bic

plot(regfit.bwd,scale="bic")

Validation Set Approach


set.seed(1)
train=sample(c(TRUE,FALSE),nrow(Hitters),rep=T)
test =(!train )
regfit.best=regsubsets(Salary~.,data=Hitters[train,],nvmax =19)
test.mat=model.matrix(Salary~.,data=Hitters[test,])
val.errors =rep(NA ,19)
for(i in 1:19){
coefi=coef(regfit.best,id=i)
pred=test.mat[,names(coefi)]%*% coefi
val.errors[i]= mean((Hitters$Salary[test]-pred)^2)
}
val.errors
## [1] 220968.0 169157.1 178518.2 163426.1 168418.1 171270.6 162377.1
## [8] 157909.3 154055.7 148162.1 151156.4 151742.5 152214.5 157358.7
## [15] 158541.4 158743.3 159972.7 159859.8 160105.6

13

which.min(val.errors)
## [1] 10
coef(regfit.best,10)
## (Intercept)
## -80.2751499
##
CHmRun
##
1.3844863

AtBat
-1.4683816
CWalks
-0.7483170

Hits
Walks
7.1625314
3.6430345
LeagueN
DivisionW
84.5576103 -53.0289658

CAtBat
-0.1855698
PutOuts
0.2381662

CHits
1.1053238

regfit.best=regsubsets (Salary~.,data=Hitters ,nvmax =19)


coef(regfit.best ,10)
##
##
##
##
##
##

(Intercept)
162.5354420
CRuns
1.4082490
Assists
0.2831680

AtBat
-2.1686501
CRBI
0.7743122

Hits
Walks
6.9180175
5.7732246
CWalks
DivisionW
-0.8308264 -112.3800575

CAtBat
-0.1300798
PutOuts
0.2973726

Cross-Validation
predict.regsubsets =function(object ,newdata ,id ,...){
form=as.formula(object$call [[2]])
mat=model.matrix(form ,newdata )
coefi=coef(object,id=id)
xvars=names(coefi)
mat[,xvars]%*% coefi
}
k=10
set.seed(1)
folds=sample(1:k,nrow(Hitters),replace =TRUE)
cv.errors=matrix(NA,k,19,dimnames=list(paste(1:k), paste(1:19)))
for(j in 1:k){
best.fit =regsubsets(Salary~.,data=Hitters[folds!=j,],nvmax =19)
for(i in 1:19) {
pred=predict(best.fit ,Hitters[folds==j,],id=i)
cv.errors [j,i]=mean( (Hitters$Salary[folds ==j]-pred)^2)
}
}
mean.cv.errors=apply(cv.errors ,2, mean)
mean.cv.errors
##
1
2
3
4
5
6
7
8
## 160093.5 140196.8 153117.0 151159.3 146841.3 138302.6 144346.2 130207.7
##
9
10
11
12
13
14
15
16
## 129459.6 125334.7 125153.8 128273.5 133461.0 133974.6 131825.7 131882.8
##
17
18
19
## 132750.9 133096.2 132804.7
14

125000 135000 145000 155000

mean.cv.errors

plot(mean.cv.errors,type="b")

10

15

Index

reg.best=regsubsets(Salary~.,data=Hitters,nvmax =19)
coef(reg.best ,11)

##
##
##
##
##
##

(Intercept)
135.7512195
CRuns
1.4553310
PutOuts
0.2894087

AtBat
-2.1277482
CRBI
0.7852528
Assists
0.2688277

Hits
6.9236994
CWalks
-0.8228559

Walks
CAtBat
5.6202755
-0.1389914
LeagueN
DivisionW
43.1116152 -111.1460252

Reference:
James, Gareth, et al. An introduction to statistical learning. New
York: springer, 2013.

15

You might also like