12 février 2021
Arthur Charpentier, « Some general thoughts on Partial Dependence Plots with correlated covariates », Freakonometrics, ID : 10670/1.ts76t5
The partial dependence plot is a nice tool to analyse the impact of some explanatory variables when using nonlinear models, such as a random forest, or some gradient boosting.The idea (in dimension 2), given a model [latex]m(x_1,x_2)[/latex] for [latex]\mathbb{E}[Y|X_1=x_1,X_2=x_2][/latex]. The partial dependence plot for variable [latex]x_1[/latex] is model [latex]m[/latex] is function [latex]p_1[/latex] defined as [latex]x_1\mapsto\mathbb{E}_{\mathbb{P}_{X_2}}[m(x_1,X_2)][/latex]. This can ...