library(ISLR)
## Warning: package 'ISLR' was built under R version 4.2.2
library(tree)
## Warning: package 'tree' was built under R version 4.2.2
library(rpart)
library(caret)
## Warning: package 'caret' was built under R version 4.2.2
## Loading required package: ggplot2
## Loading required package: lattice
library(randomForest)
## Warning: package 'randomForest' was built under R version 4.2.2
## randomForest 4.7-1.1
## Type rfNews() to see new features/changes/bug fixes.
##
## Attaching package: 'randomForest'
## The following object is masked from 'package:ggplot2':
##
## margin
library(ISLR)
3
p = seq(0, 1, 0.001)
gindex <- 2 * p * (1 - p)
clerror <- 1 - pmax(p, 1 - p)
crent <- - (p * log(p) + (1 - p) * log(1 - p))
matplot(p, cbind(gindex, clerror, crent), ylab = "gindex, clerror, crent", col = c("green", "blue", "red"))
legend('bottom', inset=.01, legend = c('gini index', 'class error', 'cross entropy'), col = c("green", "blue", "red"), pch=c(15,17,19))
8
A
set.seed(1)
train <- sample(1:nrow(Carseats), nrow(Carseats)/2)
training <- Carseats[train, ]
testing <- Carseats[-train, ]
B
tree.carseats <- tree(Sales ~ ., data = training)
summary(tree.carseats)
##
## Regression tree:
## tree(formula = Sales ~ ., data = training)
## Variables actually used in tree construction:
## [1] "ShelveLoc" "Price" "Age" "Advertising" "CompPrice"
## [6] "US"
## Number of terminal nodes: 18
## Residual mean deviance: 2.167 = 394.3 / 182
## Distribution of residuals:
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## -3.88200 -0.88200 -0.08712 0.00000 0.89590 4.09900
plot(tree.carseats)
text(tree.carseats, pretty = 0)
treecarseat.pred <- predict(tree.carseats, newdata = testing)
mean((treecarseat.pred - testing$Sales)^2)
## [1] 4.922039
The MSE is 4.922039
C
set.seed(1)
cv.carseats <- cv.tree(tree.carseats)
plot(cv.carseats$size, cv.carseats$dev, type = "b")
The optimal level is 12
prune.carseats <- prune.tree(tree.carseats, best = 12)
plot(prune.carseats)
text(prune.carseats,pretty=0)
treecarseat.pred <- predict(prune.carseats, newdata = testing)
mean((treecarseat.pred - testing$Sales)^2)
## [1] 4.966929
The MSE moved from 4.922039 to 4.918134!
D
set.seed(1)
bag.carseats <- randomForest(Sales~., data = training, mtry = 10, ntree = 551, importance = TRUE)
bagcarseat.pred <- predict(bag.carseats, newdata = testing)
mean((bagcarseat.pred - testing$Sales)^2)
## [1] 2.599099
The MSE is 2.599099!
importance(bag.carseats)
## %IncMSE IncNodePurity
## CompPrice 26.18616309 170.781666
## Income 5.25063979 90.717958
## Advertising 13.25673204 97.498810
## Population -2.14346969 58.289311
## Price 60.58241525 503.478806
## ShelveLoc 50.77308639 380.258594
## Age 19.03720001 158.282846
## Education 1.24264920 44.834257
## Urban -0.08461165 9.883299
## US 4.71515903 17.907727
Price and ChelveLoc are the highest
set.seed(1)
rando.carseats = randomForest(Sales~., data = training, mtry = 10, importance = TRUE)
randcarseat.pred = predict(rando.carseats, newdata = testing)
mean((randcarseat.pred - testing$Sales)^2)
## [1] 2.605253
The MSE is 2.605253
importance(rando.carseats)
## %IncMSE IncNodePurity
## CompPrice 24.8888481 170.182937
## Income 4.7121131 91.264880
## Advertising 12.7692401 97.164338
## Population -1.8074075 58.244596
## Price 56.3326252 502.903407
## ShelveLoc 48.8886689 380.032715
## Age 17.7275460 157.846774
## Education 0.5962186 44.598731
## Urban 0.1728373 9.822082
## US 4.2172102 18.073863
The most important are the same as above.
9 A
set.seed(1)
train <- sample(1:nrow(OJ), 800)
ojtraining <- OJ[train, ]
ojtesting <- OJ[-train, ]
B
ojtree <- tree(Purchase ~ ., data = ojtraining)
summary(ojtree)
##
## Classification tree:
## tree(formula = Purchase ~ ., data = ojtraining)
## Variables actually used in tree construction:
## [1] "LoyalCH" "PriceDiff" "SpecialCH" "ListPriceDiff"
## [5] "PctDiscMM"
## Number of terminal nodes: 9
## Residual mean deviance: 0.7432 = 587.8 / 791
## Misclassification error rate: 0.1588 = 127 / 800
The tree has 0 terminal nodes and an error rate of .1588
C
ojtree
## node), split, n, deviance, yval, (yprob)
## * denotes terminal node
##
## 1) root 800 1073.00 CH ( 0.60625 0.39375 )
## 2) LoyalCH < 0.5036 365 441.60 MM ( 0.29315 0.70685 )
## 4) LoyalCH < 0.280875 177 140.50 MM ( 0.13559 0.86441 )
## 8) LoyalCH < 0.0356415 59 10.14 MM ( 0.01695 0.98305 ) *
## 9) LoyalCH > 0.0356415 118 116.40 MM ( 0.19492 0.80508 ) *
## 5) LoyalCH > 0.280875 188 258.00 MM ( 0.44149 0.55851 )
## 10) PriceDiff < 0.05 79 84.79 MM ( 0.22785 0.77215 )
## 20) SpecialCH < 0.5 64 51.98 MM ( 0.14062 0.85938 ) *
## 21) SpecialCH > 0.5 15 20.19 CH ( 0.60000 0.40000 ) *
## 11) PriceDiff > 0.05 109 147.00 CH ( 0.59633 0.40367 ) *
## 3) LoyalCH > 0.5036 435 337.90 CH ( 0.86897 0.13103 )
## 6) LoyalCH < 0.764572 174 201.00 CH ( 0.73563 0.26437 )
## 12) ListPriceDiff < 0.235 72 99.81 MM ( 0.50000 0.50000 )
## 24) PctDiscMM < 0.196196 55 73.14 CH ( 0.61818 0.38182 ) *
## 25) PctDiscMM > 0.196196 17 12.32 MM ( 0.11765 0.88235 ) *
## 13) ListPriceDiff > 0.235 102 65.43 CH ( 0.90196 0.09804 ) *
## 7) LoyalCH > 0.764572 261 91.20 CH ( 0.95785 0.04215 ) *
Terminal node 11 the split criterion is PriceDiff > 0.05, the number of observations is 109 with a dev. of 147.
D
plot(ojtree)
text(ojtree, pretty = 0)
E
tree.pred <- predict(ojtree, ojtesting, type = "class")
table(tree.pred, ojtesting$Purchase)
##
## tree.pred CH MM
## CH 160 38
## MM 8 64
1-(160+64)/270
## [1] 0.1703704
The error rate is 17.03%
F
cv.oj <- cv.tree(ojtree, FUN = prune.misclass)
cv.oj
## $size
## [1] 9 8 7 4 2 1
##
## $dev
## [1] 150 150 149 158 172 315
##
## $k
## [1] -Inf 0.000000 3.000000 4.333333 10.500000 151.000000
##
## $method
## [1] "misclass"
##
## attr(,"class")
## [1] "prune" "tree.sequence"
G
plot(cv.oj$size, cv.oj$dev, type = "b", xlab = "Tree size", ylab = "Deviance")
H
The 7 node tree is the smallest with the lowest error rate.
I
prune.oj <- prune.misclass(ojtree, best = 7)
plot(prune.oj)
text(prune.oj, pretty = 0)
J
summary(prune.oj)
##
## Classification tree:
## snip.tree(tree = ojtree, nodes = c(4L, 10L))
## Variables actually used in tree construction:
## [1] "LoyalCH" "PriceDiff" "ListPriceDiff" "PctDiscMM"
## Number of terminal nodes: 7
## Residual mean deviance: 0.7748 = 614.4 / 793
## Misclassification error rate: 0.1625 = 130 / 800
summary(ojtree)
##
## Classification tree:
## tree(formula = Purchase ~ ., data = ojtraining)
## Variables actually used in tree construction:
## [1] "LoyalCH" "PriceDiff" "SpecialCH" "ListPriceDiff"
## [5] "PctDiscMM"
## Number of terminal nodes: 9
## Residual mean deviance: 0.7432 = 587.8 / 791
## Misclassification error rate: 0.1588 = 127 / 800
The unpruned tree has a lower misclassification rate (15.88% vs 16.25%)
K
prune.pred <- predict(prune.oj, ojtesting, type = "class")
table(prune.pred, ojtesting$Purchase)
##
## prune.pred CH MM
## CH 160 36
## MM 8 66
1-(160+66)/270
## [1] 0.162963
The error rate has gone to 16.3%