rm(list = ls()) library(openxlsx) library(MASS) data = read.xlsx("xiti_4.xlsx",sheet = 2) data fm = lm(y~x1+x2+x3+x4+x5+x6+x7,data) par(mfrow = c(2,2),mar = 0.4+c(4,4,1,1),oma = c(0,0,2,0)) a1 = boxcox(fm,lambda = seq(0,1,by = 0.1))
#λ=0.76 l = which(a1$y==max(a1$y)) a1$x[l] #得到0.76就是图像的最高点
lamb = 0.76 ylam = (data$y^lamb -1)/lamb new_data = cbind(data,ylam) fm1 = lm(ylam~x1+x2+x3+x4+x5+x6+x7,new_data ) #数据变换后拟合的模型 #计算t外部学生化残差 ei = resid(fm1) X = cbind(1,as.matrix(data[,2:8])) t = ti(ei,X)
#残差图 plot(fitted(fm1),t)
从图中看出,没有显示出明显的方差不齐的迹象
#正态性检验 shapiro.test(resid(fm1))
> shapiro.test(resid(fm1)) Shapiro-Wilk normality test data: resid(fm1) W = 0.97405, p-value = 0.748
正态性检验通过
最终的回归模型: