[Feature Request]: Add Double/Debiased Machine Learning for Dynamic Treatment Effects

[Akramerhh]

Describe the feature you want to propose or implement

Dynamic treatment effects are highly relevant in practice, where the assignment to multipe treatments over time is rather the default then the exception.

Propose a possible solution or implementation

Implement the proposed methods in:

Lewis, G., Syrgkanis, V. (2021). Double/Debiased Machine Learning for Dynamic Treatment Effects. URL https://arxiv.org/abs/2002.07285

similar to:

https://econml.azurewebsites.net/spec/estimation/dynamic_dml.html

Did you consider alternatives to the proposed solution. If yes, please describe

No response

Comments, context or references

Big fan of your package, highly appreciate your efforts to make sophisticated causal inference available in R.

[PhilippBach]

Hi @Akramerhh ,

thanks for opening this feature request and your nice feedback. We are adding this to the list of planned extension. I can't promise when we will integrate such a feature in the R package, as we are focusing more on the Python implementation right now (due to resource constraints). We plan to extend the R package too. If you want to add a new feature, you are of course welcome to contribute!

Thanks again and best,

Philipp