CausalForestDML need X to successfully run about econml HOT 2 CLOSED

The point of CausalForestDML is to fit a forest on X, which isn't possible X is None. So if all you want to do is compute an ATE, then just use LinearDML with X=None instead.

On the other hand, if you use CausalForestDML to compute a conditional average treatment effect (conditional on some actual variables in X), then that class provides both an ate_ attribute which has a doubly-robust estimation of the ATE on the training data as well as ate, ate_inference, and ate_interval methods which will estimate the ATE, confidence intervals, etc. on some other population of Xs by averaging the estimated CATE over them.

from econml.

Thanks!

from econml.