Consider the Gini index, classification error, and entropy in a simple classification setting with two classes. Create a single plot that displays each of these quantities as a function of ˆpm1. The xaxis should display ˆpm1, ranging from 0 to 1, and the y-axis should display the value of the Gini index, classification error, and entropy. Hint: In a setting with two classes, pˆm1 = 1 − pˆm2. You could make this plot by hand, but it will be much easier to make in R
pmk = seq(0, 1, 0.01)
gini.index = 2 * pmk * (1 - pmk)
classification.error = 1 - pmax(pmk, 1 - pmk)
cross.entropy = - (pmk * log(pmk) + (1 - pmk) * log(1 - pmk))
matplot(pmk, cbind(gini.index, classification.error, cross.entropy), col = c("black", "blue", "red"), pch=c(1,1,1))
legend('topright', inset=.01, legend = c('gini index', 'classification error', 'cross entropy'), col = c("black" , "blue", "red"), pch=c(1,1,1))
In the lab, a classification tree was applied to the Carseats data set after converting Sales into a qualitative response variable. Now we will seek to predict Sales using regression trees and related approaches, treating the response as a quantitative variable.
library(ISLR)
attach(Carseats)
library(rpart)
library(caret)
library(tree)
(a) Split the data set into a training set and a test set.
set.seed(1)
train = sample(1:nrow(Carseats), nrow(Carseats) / 2)
training.set = Carseats[train, ]
testing.set = Carseats[-train,]
(b) Fit a regression tree to the training set. Plot the tree, and interpret the results. What test MSE do you obtain?
tree.carseats = tree(Sales~.,data = training.set)
summary(tree.carseats)
##
## Regression tree:
## tree(formula = Sales ~ ., data = training.set)
## 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)
tree.predict = predict(tree.carseats,newdata = testing.set)
mean((tree.predict - testing.set$Sales)^2)
## [1] 4.922039
Using the regression tree, we can get MSE of 4.922039.
(c) Use cross-validation in order to determine the optimal level of tree complexity. Does pruning the tree improve the test MSE?
set.seed(1)
cv.carseats = cv.tree(tree.carseats)
plot(cv.carseats$size, cv.carseats$dev, type = "b")
prune.carseats = prune.tree(tree.carseats, best = 8)
plot(prune.carseats)
text(prune.carseats,pretty=0)
pred.cv.carseats =predict(prune.carseats, newdata= testing.set)
mean((pred.cv.carseats-testing.set$Sales)^2)
## [1] 5.113254
The test MSE for cross-validation model is 5.113254
(d) Use the bagging approach in order to analyze this data. What test MSE do you obtain? Use the importance() function to determine which variables are most important.
library(randomForest)
set.seed(1)
bagging.carseats = randomForest(Sales~.,data=training.set,mtry = 10, importance = TRUE)
pred.bagging.carseats = predict(bagging.carseats,newdata=testing.set)
mean((pred.bagging.carseats-testing.set$Sales)^2)
## [1] 2.605253
importance(bagging.carseats)
## %IncMSE IncNodePurity
## CompPrice 24.8888481 170.182937
## Income 4.7121131 91.264880
## Advertising 12.7692401 97.164338
## Population -1.8074075 58.244596
## Price 56.3326252 502.903407
## ShelveLoc 48.8886689 380.032715
## Age 17.7275460 157.846774
## Education 0.5962186 44.598731
## Urban 0.1728373 9.822082
## US 4.2172102 18.073863
varImpPlot(bagging.carseats)
The MSE for bagging approach is 2.605253. We can see that price is the most important predictor in this model.
(e) Use random forests to analyze this data. What test MSE do you obtain? Use the importance() function to determine which variables are most important. Describe the effect of m, the number of variables considered at each split, on the error rate obtained.
set.seed(1)
rf.carseats = randomForest(Sales~.,data=training.set,mtry = 3, importance = TRUE)
pred.rf = predict(rf.carseats,newdata=testing.set)
mean((pred.rf-testing.set$Sales)^2)
## [1] 2.960559
importance(rf.carseats)
## %IncMSE IncNodePurity
## CompPrice 14.8840765 158.82956
## Income 4.3293950 125.64850
## Advertising 8.2215192 107.51700
## Population -0.9488134 97.06024
## Price 34.9793386 385.93142
## ShelveLoc 34.9248499 298.54210
## Age 14.3055912 178.42061
## Education 1.3117842 70.49202
## Urban -1.2680807 17.39986
## US 6.1139696 33.98963
varImpPlot(rf.carseats)
We can see that MSE for random forest model is 2.960559. And similar to bagging model this model also has price predictor as most important predictor
This problem involves the OJ data set which is part of the ISLR package.
library(ISLR)
attach(OJ)
library(tree)
(a) Create a training set containing a random sample of 800 observations, and a test set containing the remaining observations.
train = sample(dim(OJ)[1], 800)
training.set = OJ[train, ]
testing.set = OJ[-train, ]
(b) Fit a tree to the training data, with Purchase as the responseand the other variables as predictors. Use the summary() function to produce summary statistics about the tree, and describe the results obtained. What is the training error rate? How many terminal nodes does the tree have?
library(tree)
OJ.tree = tree(Purchase ~ ., data = training.set)
summary(OJ.tree)
##
## Classification tree:
## tree(formula = Purchase ~ ., data = training.set)
## Variables actually used in tree construction:
## [1] "LoyalCH" "SalePriceMM" "PriceDiff"
## Number of terminal nodes: 6
## Residual mean deviance: 0.7797 = 619.1 / 794
## Misclassification error rate: 0.1812 = 145 / 800
As you can see there are 7 terminal nodes. And the MSE is 0.1588
(c) Type in the name of the tree object in order to get a detailed text output. Pick one of the terminal nodes, and interpret the information displayed.
OJ.tree
## node), split, n, deviance, yval, (yprob)
## * denotes terminal node
##
## 1) root 800 1074.00 CH ( 0.60375 0.39625 )
## 2) LoyalCH < 0.482935 310 331.20 MM ( 0.22581 0.77419 )
## 4) LoyalCH < 0.0616725 75 18.44 MM ( 0.02667 0.97333 ) *
## 5) LoyalCH > 0.0616725 235 282.70 MM ( 0.28936 0.71064 )
## 10) SalePriceMM < 2.04 132 128.10 MM ( 0.18939 0.81061 ) *
## 11) SalePriceMM > 2.04 103 140.00 MM ( 0.41748 0.58252 ) *
## 3) LoyalCH > 0.482935 490 426.20 CH ( 0.84286 0.15714 )
## 6) LoyalCH < 0.764572 232 277.10 CH ( 0.71552 0.28448 )
## 12) PriceDiff < 0.145 88 121.90 MM ( 0.48864 0.51136 ) *
## 13) PriceDiff > 0.145 144 119.60 CH ( 0.85417 0.14583 ) *
## 7) LoyalCH > 0.764572 258 90.94 CH ( 0.95736 0.04264 ) *
We can pick \(25)\) node as it is the terminal node. we can see that there are 22 observation and PctDiscMM > 0.196197. Also \(25\) is considered as “CH”
(d) Create a plot of the tree, and interpret the results.
plot(OJ.tree)
text(OJ.tree, pretty = 0)
We can see that if LoyalCH<0.282272 then it is classified as “MM” while 0.753545<LoyalCH then it is classified as “CH”.
(e) Predict the response on the test data, and produce a confusion matrix comparing the test labels to the predicted test labels. What is the test error rate?
OJ.pred = predict(OJ.tree, testing.set, type = "class")
test.table<-table(testing.set$Purchase, OJ.pred)
(test.table[2,1] + test.table[2,1]) / sum(test.table)
## [1] 0.1111111
We can see that it predicts error rate of 0.1925926
(f) Apply the cv.tree() function to the training set in order to determine the optimal tree size.
cv.OJ = cv.tree(OJ.tree, FUN = prune.tree)
(g) Produce a plot with tree size on the x-axis and cross-validated classification error rate on the y-axis.
plot(cv.OJ$size, cv.OJ$dev, type = "b",)
(h) Which tree size corresponds to the lowest cross-validated classification error rate?
We can see that at tree size of 6 we get the lowest cv error rate.
(i) Produce a pruned tree corresponding to the optimal tree size obtained using cross-validation. If cross-validation does not lead to selection of a pruned tree, then create a pruned tree with five terminal nodes.
OJ.pruned = prune.tree(OJ.tree, best = cv.OJ$size[which.min(cv.OJ$dev)])
(j) Compare the training error rates between the pruned and unpruned trees. Which is higher?
summary(OJ.pruned)
##
## Classification tree:
## snip.tree(tree = OJ.tree, nodes = 5L)
## Variables actually used in tree construction:
## [1] "LoyalCH" "PriceDiff"
## Number of terminal nodes: 5
## Residual mean deviance: 0.7971 = 633.7 / 795
## Misclassification error rate: 0.1812 = 145 / 800
We can see that pruned tree has higher training error rates compared to the unpruned ones which had error rates of 0.1588
(k) Compare the test error rates between the pruned and unpruned trees. Which is higher?
pred.unpruned = predict(OJ.tree, testing.set, type = "class")
mse.unpruned = sum(testing.set$Purchase != pred.unpruned)
mse.unpruned/length(pred.unpruned)
## [1] 0.2037037
pred.pruned = predict(OJ.pruned, testing.set, type = "class")
mse.pruned = sum(testing.set$Purchase != pred.pruned)
mse.pruned/length(pred.pruned)
## [1] 0.2037037
We can see that pruned tree has higher MSE values of .188889>.162963 in testing set also.