How to use PolicyTree from a CausalForest to predict treatment of new unseen data?

Open titubs opened this issue 2 years ago • 0 comments

@kbattocchi Hi Keith! I hope you had a great start into 2023. I wanted to ask you if I am able to use a Policy Tree and score new data on it? I.e. predict that is the optimal treatment for a given instance.

My model choice was a CausalForest BUT when I follow the "Policy Learning" section here https://github.com/microsoft/EconML it tells me that CausalForest does not have any predict(X) option. Is that true or is there any other trick? Does predict(X) ONLY come with a subset of learners such as DR, etc?

Feb 14 '23 20:02 titubs