Trees, forests, and impurity-based variable importance

Abstract

Tree ensemble methods such as random forests [Breiman, 2001] are very popular to handle high-dimensional tabular data sets, notably because of their good predictive accuracy. However, when machine learning is used for decision-making problems, settling for the best predictive procedures may not be reasonable since enlightened decisions require an in-depth comprehension of the algorithm prediction process. Unfortunately, random forests are not intrinsically interpretable since their prediction results from averaging several hundreds of decision trees. A classic approach to gain knowledge on this so-called black-box algorithm is to compute variable importances, that are employed to assess the predictive impact of each input variable. Variable importances are then used to rank or select variables and thus play a great role in data analysis. Nevertheless, there is no justification to use random forest variable importances in such way: we do not even know what these quantities estimate. In this paper, we analyze one of the two well-known random forest variable importances, the Mean Decrease Impurity (MDI). We prove that if input variables are independent and in absence of interactions, MDI provides a variance decomposition of the output, where the contribution of each variable is clearly identified. We also study models exhibiting dependence between input variables or interaction, for which the variable importance is intrinsically ill-defined. Our analysis shows that there may exist some benefits to use a forest compared to a single tree.

References in corpus (2)

• Random Forests: some methodological insights
• A Debiased MDI Feature Importance Measure for Random Forests

Cited by in corpus (9)

• Impact of Accuracy on Model Interpretations
• Asymptotic Properties of High-Dimensional Random Forests
• ControlBurn: Feature Selection by Sparse Forests
• Learning Real Estate Automated Valuation Models from Heterogeneous Data Sources
• From global to local MDI variable importances for random forests and when they are Shapley values
• A cautionary tale on fitting decision trees to data from additive models: generalization lower bounds
• Nonparametric Variable Screening with Optimal Decision Stumps
• Evolutionary Optimization of High-Coverage Budgeted Classifiers
• Bridging Breiman's Brook: From Algorithmic Modeling to Statistical Learning