R语言 - 决策树 - R语言教程-OStack程序员社区-中国程序员成长平台

OStack程序员社区-中国程序员成长平台 › 门户 › 编程› R语言›R语言教程

R语言 - 决策树

原作者: [db:作者] 来自: [db:来源] 收藏邀请

#将DTdata.csv中的数据带入play_decision变量中，有header，分隔符为,
> play_decision <- read.table("DTdata.csv",header = TRUE,sep = ",")

#查看数据
> play_decision

 Play Outlook Temperature Humidity Wind

1 yes rainy cool normal FALSE

2 no rainy cool normal TRUE

3 yes overcast hot high FALSE

4 no sunny mild high FALSE

5 yes rainy cool normal FALSE

6 yes sunny cool normal FALSE

7 yes rainy cool normal FALSE

8 yes sunny hot normal FALSE

9 yes overcast mild high TRUE

10 no sunny mild high TRUE

#用rpart函数生成决策树模型，基于4种属性来预测Play属性，method="class"表示构建分类树，data指定了包含属性的数据帧，control控制
树的增长minsplit=1要求每个节点在尝试分裂前必须要有至少一个观测值
> fit <- rpart(Play ~ Outlook + Temperature + Humidity + Wind , method = "class" , data = play_decision , 

control = rpart.control(minsplit = 1) , parms = list(split="information"))

#观察rpart生成的决策树模型概览
> summary(fit)


Call:

rpart(formula = Play ~ Outlook + Temperature + Humidity + Wind, 
 data = play_decision, method = "class", parms = list(split = "information"), 
 control = rpart.control(minsplit = 1))
 n= 10 

 CP nsplit rel error xerror xstd

1 0.3333333 0 1 1.000000 0.4830459

2 0.0100000 3 0 1.666667 0.5270463


Variable importance
 Wind Outlook Temperature 
 51 29 20 


Node number 1: 10 observations, complexity param=0.3333333
 predicted class=yes expected loss=0.3 P(node) =1
 class counts: 3 7
 probabilities: 0.300 0.700 
 left son=2 (3 obs) right son=3 (7 obs)
 Primary splits:
 Temperature splits as RRL, improve=1.3282860, (0 missing)
 Wind < 0.5 to the right, improve=1.3282860, (0 missing)
 Outlook splits as RLL, improve=0.8161371, (0 missing)
 Humidity splits as LR, improve=0.6326870, (0 missing)
 Surrogate splits:
 Wind < 0.5 to the right, agree=0.8, adj=0.333, (0 split)


Node number 2: 3 observations, complexity param=0.3333333
 predicted class=no expected loss=0.3333333 P(node) =0.3
 class counts: 2 1
 probabilities: 0.667 0.333 
 left son=4 (2 obs) right son=5 (1 obs)
 Primary splits:
 Outlook splits as R-L, improve=1.9095430, (0 missing)
 Wind < 0.5 to the left, improve=0.5232481, (0 missing)


Node number 3: 7 observations, complexity param=0.3333333
 predicted class=yes expected loss=0.1428571 P(node) =0.7
 class counts: 1 6
 probabilities: 0.143 0.857 
 left son=6 (1 obs) right son=7 (6 obs)
 Primary splits:
 Wind < 0.5 to the right, improve=2.8708140, (0 missing)
 Outlook splits as RLR, improve=0.6214736, (0 missing)
 Temperature splits as LR-, improve=0.3688021, (0 missing)
 Humidity splits as RL, improve=0.1674470, (0 missing)


Node number 4: 2 observations
 predicted class=no expected loss=0 P(node) =0.2
 class counts: 2 0
 probabilities: 1.000 0.000 


Node number 5: 1 observations
 predicted class=yes expected loss=0 P(node) =0.1
 class counts: 0 1
 probabilities: 0.000 1.000 


Node number 6: 1 observations
 predicted class=no expected loss=0 P(node) =0.1
 class counts: 1 0
 probabilities: 1.000 0.000 


Node number 7: 6 observations
 predicted class=yes expected loss=0 P(node) =0.6
 class counts: 0 6
 probabilities: 0.000 1.000 

#将已生成的决策树可视化
> rpart.plot(fit , type = 4 , extra = 1)

#建立新的数据帧
> newdata <- data.frame(Outlook="rainy",Temperature="mild",Humidity="high",Wind=FALSE)

#查看新建立的数据帧
> newdata

#利用predict函数预测newdata数据帧的属性是否会Play，参数type表示预测值类型
> predict(fit,newdata=newdata,type="prob")

 no yes

1 1 0

或

> predict(fit,newdata=newdata,type="class")

 1 

no 

Levels: no yes

#结果显示两种预测值类型结果都显示no