R 《回归分析与线性统计模型》page120,4.3
#P120习题4.3 rm(list = ls()) A = read.xlsx("xiti_4.xlsx",sheet = 3) names(A) = c("ord","Y","K","L") attach(A) fm = lm(Y~log(K)+log(L))#线性回归模型 ei = resid(fm) X = cbind(1,as.matrix(A[,3:4])) t = ti(ei,X) #外部学生化残差 plot(fitted(fm),t) #绘制残差图
从残差图中看出来,方差不齐
a1 = boxcox(fm,lambda = seq(0,1,by=0.1))
从图像中看出,λ可取0,即进行对数变换
#进行对数变换 lm.log = lm(log(Y)~log(L)+log(K)) coef(lm.log) summary(lm.log) detach(A)
> summary(lm.log) Call: lm(formula = log(Y) ~ log(L) + log(K)) Residuals: Min 1Q Median 3Q Max -1.7251 -0.1764 -0.0059 0.1707 1.3035 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 0.38004 0.26873 1.414 0.166 log(L) 0.05699 0.04471 1.275 0.211 log(K) 0.93065 0.04131 22.526 <2e-16 *** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 0.441 on 35 degrees of freedom Multiple R-squared: 0.944, Adjusted R-squared: 0.9408 F-statistic: 295 on 2 and 35 DF, p-value: < 2.2e-16