Paul Zivich

Possible new branch: meta-analysis tools. For missing data: https://ebmh.bmj.com/content/17/3/85

Currently `AIPTW` only returns RD and RR. `TMLE` returns those and OR as well. I should add support for OR with `AIPTW` (even though I am not a huge fan...

Currently you need to set the `np.random.seed` outside of the function for reproducibility (which isn't good). I should use a similar RandomState approach that the cross-fit estimators use

Update documentation (and possibly re-organize)

2

I wrote most of the ReadTheDocs documentation 2-3 years ago now. It is dated (and my understanding has expanded), so I should go back and review everything after the v0.9.0...

Other helpful resources

1

This Issue thread is meant to contain a collection of potentially helpful outside resources that don't belong with a specific function. It may also include broad future additions Nice causal...

Update the `ValueError` checks in the `test/*` to check that the correct `ValueError` is being thrown. The new `tests/test_superlearner` demonstrates this (and is much better as a check)

Add cross-fitting estimators

3

Part of v0.9.0 update. All cross-fit estimators I would like to add - [x] AIPTW - [ ] AIPTW with censored data - [x] TMLE - [ ] TMLE with...

Currently functionality is not supported for two treatments, like looking at `A` and `B`. For binary treatments, this consists of 4 different options. It may be worth considering the addition...

When the MSM are fit in the `IPTW.fit()` statement, it is possible for a `PerfectSeparationError` to occur. This is more of a concern for the linear-binomial models (as opposed to...

A few items have come up regarding `IPTW`. One additional diagnostic plot to add is a weighted histogram for continuous variables. This way it can be visually assessed whether the...