Keith Battocchi

Could you provide the full stack trace, as well as the output of `pip list`, please?

Could you please provide a full reproduction of this issue (perhaps with a simple dummy dataset), including the stack trace? I'd like to take a closer look.

We don't include our tests module in our published package, it's mainly used for internal testing, so this is by design. However, I see that we do use this in...

I think using a wrapper is the best workaround for now, but we'll consider adding direct support - see https://github.com/microsoft/EconML/issues/543.

@MasaAsami is correct that you can gain some control via preprocessing in a pipeline, but be aware that the features always include the columns of both X and W, and...

Thanks for reporting this issue. What versions of econml and dowhy do you have installed?

Sorry for the unhelpful error message; the issue ultimately is that you need to call `fit` through the dowhy wrapper first before calling `refute_estimate`, because that will not only fit...

This is a great suggestion, thanks.

We do not currently have any plans to support this, but it's on our radar. Could you expand on your use case for this a bit? It looks like LightGBM...

The models themselves are stored in the `propensity_models` attribute of the estimator; we calculate but do not store the propensity scores as part of the fitting process; we do not...