py-why / py-why/EconML

Bootstrap tests are flaky, unprincipled

Open
#127 0 comments 0 reactions 2 assignees View on GitHub

@vasilismsr is already working on this.

Since Oct 31, 2019.

Dominant language
Jupyter Notebook
Stars
4.8k
Forks
827
PR merge metrics
No merged PRs in 30d

Description

The test_bootstrap.TestBootstrap.test_internal_options test fails sporadically due to its random data generation.

We could fix a single random seed so that the test is deterministic. However, the test itself is also bad for a few reasons:

  1. It tests that the estimated effect is "usually" within the lower and upper bounds of the confidence interval; we should also be testing that the ground truth is within the bounds
  2. It is not averaging over several instances of the problem, but over several X, T1, T2 tuples for a single instance.
  3. The numerical values were set arbitrarily so that the test usually passes; they should instead be set to a number grounded in theory.

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.