Michael Mayer

@kiparkmoon not exactly. Lightgbm and XGBoost contain TreeSHAP as part of the library. XGBoost additionally provides SHAP interactions. It would be fantastic to have SHAP interactions also as part of...

In Python, SHAP interactions for lgb are calculated outside the Lightgbm library, namely directly in {shap}. The aim of this issue is to bring it into lightgbm, so that it...

Full support for this issue, especially since the implementation of KernelSHAP is very neat, and KernelSHAP has better properties than the permutation explainer when the number of features p is...

This is the current basic call: ```py import numpy as np import polars as pl from sklearn.linear_model import LinearRegression from model_diagnostics.xai import plot_permutation_importance rng = np.random.default_rng(1) n = 1000 X...

> The failing test is in the python 3.9 env with numpy 1.22.0 polars 1.0.0 scipy 1.10.0 pandas 1.5.3 pyarrow 11.0.0 > > Could you check if increasing one of...

I have modified these aspects in the main functionality: - `compute_permutation_importance()` now returns both score differences and score ratios - Instead of standard deviations, the function returns standard errors -...

I see. It was a lazy/naive way to avoid breaking a dependency that uses "value" instead of the new "_value".

Same here.

Solves https://github.com/liuyanguu/SHAPforxgboost/issues/33 I am using a slightly simpler switch than in "shapviz" in order to keep the code clean: If lgb has version >= 4, then use the new `predict()`,...

Great discussion! And really helpful code of @thomasjpfan My typical workflow with random forests is quite as you describe: 1. Tune 1-2 hyper-parameters (mtry, max_depth) of the random forest via...