1 读入数据集

# 导入数据

setwd("C:\\Users\\****\\Documents\\ML\\day3\\")dataset=read.csv('50_Startups.csv')head(dataset)R.D.SpendAdministrationMarketing.SpendStateProfit1165349.2136897.80471784.1NewYork192261.82162597.7151377.59443898.5California191792.13153441.5101145.55407934.5Florida191050.44144372.4118671.85383199.6NewYork182902.05142107.391391.77366168.4Florida166187.96131876.999814.71362861.4NewYork156991.1

2 数据预处理

#虚拟变量dataset$State=factor(dataset$State,levels=c('NewYork','California','Florida'),labels=c(1,2,3))

3 训练集和测试集

将数据按照4:1拆分,每一组分别包含自变量和因变量

#install.packages('caTools')library(caTools)set.seed(123)split=sample.split(dataset$Profit,SplitRatio=0.8)training_set=subset(dataset,split==TRUE)#多自变量test_set=subset(dataset,split==FALSE)#单因变量dim(training_set)[1]405dim(test_set)[1]105

4 模型拟合及预测

通过训练集进行模型拟合得到曲线,然后将测试集的X_test带入曲线中,得到预测结果y_pred,最后将预测结果y_pred与测试集中的y_test进行比较,确定预测是否准确。

4.1 多重线性回归

regres=lm(formula=Profit~R.D.Spend+Administration+Marketing.Spend+State,data=training_set)summary(regres)Coefficients:EstimateStd.ErrortvaluePr(>|t|)(Intercept)4.965e+047.637e+036.5011.94e-07***R.D.Spend7.986e-015.604e-0214.2516.70e-16***Administration-2.942e-025.828e-02-0.5050.617Marketing.Spend3.268e-022.127e-021.5370.134State21.213e+023.751e+030.0320.974State32.376e+024.127e+030.0580.954---Signif.codes:0‘***’0.001‘**’0.01‘*’0.05‘.’0.1‘’1

4.2 进行逐步回归分析

regres.step<-step(regres)Step:AIC=735.89Profit~R.D.Spend+Marketing.SpendDfSumofSqRSSAIC<none>3.3627e+09735.89-Marketing.Spend13.1338e+083.6761e+09737.45-R.D.Spend12.3344e+102.6706e+10816.77regres2=lm(formula=Profit~R.D.Spend+Marketing.Spend,data=training_set)

4.3 预测结果

y_pred=predict(regres2,newdata=test_set)y_pred4581116202124173687.21171299.96160499.08134783.16145873.04114467.75117025.30110369.71313298447.3997668.22test_set$Profit[1]182901.99166187.94155752.60146121.95129917.04122776.86118474.03108733.99[9]99937.5997483.56

4.4 结果可视化

plot(test_set$Profit,col="red")points(y_pred,col="blue")

“R语言多元线性回归实例分析”的内容就介绍到这里了,感谢大家的阅读。如果想了解更多行业相关的知识可以关注本站网站,小编将为大家输出更多高质量的实用文章!