In this essay I talked about the challenge of separating outliers from extremes. While outliers introduce disturbing noise in your data, extremes teach us valuable lessons about all kinds of edge cases. In this context I compared various robust regression methods and implemented them in a ready to use framework.
Read more about my findings on Medium: https://jakobpoerschmann.medium.com/stop-dropping-outliers-3-upgrades-that-prepare-your-linear-regression-for-the-real-world-37693a5fc0bb