R语言 决策树 Bagging 随机森林 Random Forest 随机森林变

一、回归

二、对回归准确性进行检验

代码：

#加载包

library (tidyverse)#数据预处理

library(tidytext)#分组排序

library(glmnet)# Lasso & Ridgelibrary ( rpart)#决策树

library(ipred)# Bagging

library( randomForest)#随机森林

#、加载数据

data( 'mtcars ' )

代码：

#数据预处理

data = mtcars %>%

mutate(vs = factor(vs , levels = c(0,1), labels = c( 'V-shaped

, 'straight" )) ,

am = factor(am, levels = c(0,1), labels = c ( ' automatic ', 'manual ' )))

x = model.matrix(mpg~. ,data = data)[ ,-1]#哑元变量转换

y = data$mpg

data = data.frame (mpg = y , x)

#数据集划分

set .seed(1)

train_id = sample(1:nrow(data),0.7*nrow(data))train = data[train_id , ]

test = data[-train_id, ]

#训练模型set.seed(1)

linear_reg = lm(mpg~. ,data = train)

stepwise = step(linear_reg,direction = 'both ' ,trace = 0)cv .ridge = cv.glmnet(x = as.matrix (train[ ,-1])，

y = train$mpg,

family = 'gaussian ' ,alpha = 0,

nfolds = 5)

cv.lasso = cv.glmnet(x = as.matrix (train[ ,-1])，

y = train$mpg,

family = 'gaussian ' ,alpha = 1,

nfolds = 5)

tree = rpart(mpg~. ,data = train)

bag_tree = bagging(mpg~. , data = train)

rf = randomForest(mpg~. , data = train , importance = T)

6随机森林变量重要性

#变量重要性

varImpPlot(rf ,main = 'Variable Importance in Random Forest' )

。。。。。。。。。。。。。。。。。。。。。。。

最后：