Robust Regression Analysis in R
Arndt Regorz, Dipl. Kfm. & M.Sc. Psychology, 09/30/2021
(Note: When you click on this video you are using a service offered by YouTube.)
The following annotated code runs a robust regression in R
You will need the following R packages, each of which must be installed once before use, e.g. install.packages("robustbase"):
- robustbase
- olsrr
R code
library(robustbase)
# Dataframe from package robustbase
head(CrohnD)
# Ordinary least squares regression
fit.reg <- lm(nrAdvE ~ BMI + height, data= CrohnD)
summary(fit.reg)
# Robust regression
# IMPORTANT: N should be >= 100!
fit.rob <- lmrob(nrAdvE ~ BMI + height, data= CrohnD)
summary(fit.rob)
# Outlier analysis
library(olsrr)
ols_plot_resid_lev(fit.reg)
Additional information about the MM-estimator:
The MM-estimator, on which the robust regression in the robustbase package is based, was introduced in this paper:
Yohai, V. J. (1987). High breakdown-point and high efficiency robust estimates for regression. The Annals of Statistics, 15(2), 642-656. https://www.jstor.org/stable/2241331