AndrewjSage/RFLOWESS: A robust approach to random forest regression
This package implements the LOWESS-RF method described by Sage, Genschel, and Nettleton. Random forest prediction weights are iteratively adjusted so that outlying cases are downweighted, resulting in a more robust prediction. Outlying cases are identified using a residual analysis. This approach is motivated by Cleveland's (1979) locally weighted regression and scatterplot smoothing technique.
Getting started
Browse package contents
Package details |
|
|---|---|
| Author | Andrew Sage |
| Maintainer | Andrew Sage <ajsage@iastete.edu> |
| License | What license is it under? |
| Version | 1.0.1 |
| Package repository | View on GitHub |
| Installation |
Install the latest version of this package by entering the following in R:
|
