Showing stat.MLShow all
2 papers · 1 filter
stat.ML2026
Just add noise: Debiasing tree-based variable importance in mixed data
Jiahe Li, Omar Melikechi
Variable importance scores from tree-based methods such as random forests favor continuous predictors over categorical ones. We present a theoretical analysis of this bias and prop…
stat.ML2024
Nonparametric IPSS: Fast, flexible feature selection with false discovery control
Omar Melikechi, David B. Dunson, Jeffrey W. Miller
Feature selection is a critical task in machine learning and statistics. However, existing feature selection methods either (i) rely on parametric methods such as linear or general…