SHAP values and treatment effect values for causal forest

[abdollahpouri]

Hello,
I have been following one of the examples (here) and I see the meta learners have a . get_shap_values and plot_shap_values methods for SHAP values. Also, I think they have a fit_predict method to get the individual treatment effects. But I don't see the same methods for the causal forest approach. Can I ask how we can get those for the causal forest?

[alexander-pv]

Hi, CausalTreeRegressor and CausalRandomForestRegressor in #522 could be interpreted directly via shap (shap.TreeExplainer) with minor package update. See example.

I could open new PR for this feature and PR for causalml support in shap but #522 is still open. So we need some time to get it reviewed and merged.

[alexander-pv]

Hi, the related PR #536 is open now.