O'Reilly logo

The R Book by Michael J. Crawley

Stay ahead with the world's most comprehensive technology and business learning platform.

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, tutorials, and more.

Start Free Trial

No credit card required

Robust Fitting of Linear Models

This uses the rlm function from the MASS library. The function allows one to fit a linear model by robust regression using an M estimator, allowing robust inference for parameters and robust model selection. The robust fit is minimally influenced by outliers in the response variable, in the explanatory variable(s) or in both.

robreg<-read.table("c:\\temp\\robreg.txt",header=T)
attach(robreg)
names(robreg)

[1] "amount" "rate"

plot(amount,rate,pch=16)
abline(lm(rate~amount))
summary(lm(rate~amount))

Coefficients:
             Estimate Std. Error t value   Pr(>|t|)
(Intercept)   12.1064     1.4439   8.385   1.52e-05  ***
     amount   -1.0634     0.2552  -4.166    0.00243   **

Residual standard error: 2.851 on 9 degrees of freedom
Multiple R-Squared: 0.6585, Adjusted R-squared: 0.6206
F-statistic: 17.36 on 1 and 9 DF, p-value: 0.002425

We can leave out the maximum and minimum values of the response or explanatory variables separately or together using subset to test for influence:

summary(lm(rate~amount,subset=(rate<max(rate))))

images

Coefficients:

              Estimate   Std. Error   t value   Pr(>|t|)
(Intercept)    10.8163       1.0682    10.126   7.73e-06 ***
amount         -0.9201       0.1811    -5.081   0.000952 ***

Residual standard error: 1.964 on 8 degrees of freedom
Multiple R-Squared: 0.7634, Adjusted R-squared: 0.7339
F-statistic: 25.82 on 1 and 8 DF, p-value: 0.000952

The intercept is lower by more than 1.0 and the slope is shallower ...

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, interactive tutorials, and more.

Start Free Trial

No credit card required