R语言 决策树

#决策树 分类因变量

install.packages("rpart")

library(rpart)

fit=rpart(Species~.,method = "class",iris)  #class表示分类树，利用iris数据

fit

plot(fit)    #画出决策树

text(fit)    #树上添加文字

install.packages("rpart.plot")  #上图不好看，进一步画图

library(rpart.plot)

rpart.plot(fit)   #图形美观 标注概率

printcp(fit)  #复杂度参数

test1=data.frame(iris[-5])

pre=predict(fit,test1,type = "class")

table(pre,iris$Species)           #混淆矩阵对比结果

 

#决策树 数值型因变量 回归树

data(airquality)  #R语言自带数据

air=airquality

air=na.omit(air)  #排除缺失值

head(air)   #显示前6行

fit1=rpart(Ozone~.,method="anova",air)   #anova是回归树

rpart.plot(fit1)  #树叶上显示的是中位数

printcp(fit1)  ##复杂度参数

pfit=prune(fit1,cp=0.018)  #设定cp对决策树剪枝，避免过拟合导致泛化能力降低

rpart.plot(pfit)   #剪枝后对决策树

 

 

 

软件操作结果

> install.packages("rpart")

trying URL 'https://cran.rstudio.com/bin/windows/contrib/3.5/rpart_4.1-13.zip'

Content type 'application/zip' length 950799 bytes (928 KB)

downloaded 928 KB

 

package ‘rpart’ successfully unpacked and MD5 sums checked

 

The downloaded binary packages are in

      C:\Users\L3M309NJSJ\AppData\Local\Temp\Rtmp8aeHh4\downloaded_packages

> library(rpart)

Warning message:

程辑包‘rpart’是用R版本3.5.3 来建造的

> fit=rpart(Species~.,method = "class",iris)

> fit

n= 150

 

node), split, n, loss, yval, (yprob)

      * denotes terminal node

 

1) root 150 100 setosa (0.33333333 0.33333333 0.33333333) 

  2) Petal.Length< 2.45 50   0 setosa (1.00000000 0.00000000 0.00000000) *

  3) Petal.Length>=2.45 100  50 versicolor (0.00000000 0.50000000 0.50000000) 

    6) Petal.Width< 1.75 54   5 versicolor (0.00000000 0.90740741 0.09259259) *

    7) Petal.Width>=1.75 46   1 virginica (0.00000000 0.02173913 0.97826087) *

> plot(fit)

> text(fit)

 

> install.packages("rpart.plot")

trying URL 'https://cran.rstudio.com/bin/windows/contrib/3.5/rpart.plot_3.0.6.zip'

Content type 'application/zip' length 1064991 bytes (1.0 MB)

downloaded 1.0 MB

 

package ‘rpart.plot’ successfully unpacked and MD5 sums checked

 

The downloaded binary packages are in

      C:\Users\L3M309NJSJ\AppData\Local\Temp\Rtmp8aeHh4\downloaded_packages

> library(rpart.plot)

Warning message:

程辑包‘rpart.plot’是用R版本3.5.3 来建造的

> rpart.plot(fit)

> printcp(fit)

 

Classification tree:

rpart(formula = Species ~ ., data = iris, method = "class")

 

Variables actually used in tree construction:

[1] Petal.Length Petal.Width

 

Root node error: 100/150 = 0.66667

 

n= 150

 

    CP nsplit rel error xerror     xstd

1 0.50      0      1.00   1.19 0.049592

2 0.44      1      0.50   0.69 0.061041

3 0.01      2      0.06   0.08 0.027520

> test1=data.frame(iris[-5])

> pre=predict(fit,test1,type = "class")

> table(pre,iris$Species)

           

pre          setosa versicolor virginica

  setosa          50          0         0

  versicolor      0         49         5

  virginica       0          1        45

> #决策树 数值型因变量 回归树

> data(airquality)   #R语言自带数据

> air=airquality

> air=na.omit(air)  #排除缺失值

> head(air)

  Ozone Solar.R Wind Temp Month Day

1    41     190  7.4   67     5   1

2    36     118  8.0   72     5   2

3    12     149 12.6   74     5   3

4    18     313 11.5   62     5   4

7    23     299  8.6   65     5   7

8    19      99 13.8   59     5   8

> fit1=rpart(Ozone~.,method="anova",air)

> rpart.plot(fit1)

> printcp(fit1)

 

Regression tree:

rpart(formula = Ozone ~ ., data = air, method = "anova")

 

Variables actually used in tree construction:

[1] Solar.R Temp    Wind  

 

Root node error: 121802/111 = 1097.3

 

n= 111

 

        CP nsplit rel error  xerror    xstd

1 0.484386      0   1.00000 1.01597 0.17257

2 0.097983      1   0.51561 0.55327 0.18254

3 0.057062      2   0.41763 0.60577 0.18874

4 0.020210      3   0.36057 0.50724 0.15448

5 0.018716      4   0.34036 0.50462 0.14075

6 0.017460      5   0.32164 0.48447 0.14038

7 0.010000      6   0.30418 0.48333 0.14044

> pfit=prune(fit1,cp=0.018)

> rpart.plot(pfit)