FURP-Driven Literature Review IV
Ch'i YU Lv3

Petković, M., Kocev, D. & Džeroski, S. Feature ranking for multi-target regression. Mach Learn 109, 1179–1204 (2020). https://doi.org/10.1007/s10994-019-05829-8cement) Programme.

Feature ranking for multi-target regression

Publisher:

Springer Link, Special Issue of the Discovery Science 2017

Authors:

Matej Petokovic, Dragi Kocev, Saso Dzeroski

Index Terms - Feature ranking - Multi target regression - Tree based methods - Relief

DOI: 10.1007/s10994-019-05829-8

Background(Key Point):

Multi-target regression(MTR), which concerns problems with multiple continuous dependent/target variables, is lack of studies in performing feature ranking in the context of MTR so far.

Methodology:

Two groups of feature ranking scores for MTR: scores(Symbolic, Genie3, Random Forest score) based on ensembles(bagging, random forests, extra tress) and a score derived as an extension of the RReliefF method was studied.

A generic data-transformation approach to MTR feature ranking and thus have two version of each score.

Key Findings:

The results identify the parameters that influence the quality of the rankings, reveal that both groups of methods produce relevant feature rankings, and show that the Symbolic and Genie3 score, coupled with random forest ensembles, yield the best rankings.s