enabled type checking for io_pymc3 input #1629
Conversation
Codecov Report
@@ Coverage Diff @@
## main #1629 +/- ##
=======================================
Coverage 90.92% 90.92%
=======================================
Files 108 108
Lines 11663 11665 +2
=======================================
+ Hits 10604 10606 +2
Misses 1059 1059
Continue to review full report at Codecov.
|
CHANGELOG.md
Outdated
| @@ -25,6 +25,7 @@ | |||
| * Fix `ess/rhat` plots in `plot_forest` ([1606](https://github.com/arviz-devs/arviz/pull/1606)) | |||
| * Fix `from_numpyro` crash when importing model with `thinning=x` for `x > 1` ([1619](https://github.com/arviz-devs/arviz/pull/1619)) | |||
| * Upload updated mypy.ini in ci if mypy copilot fails ([1624](https://github.com/arviz-devs/arviz/pull/1624)) | |||
| * Enforced using `pymc3.backends.base.MultiTrace` as the `trace` value for `io_pymc3` ([1629](https://github.com/arviz-devs/arviz/pull/1629)) | |||
There was a problem hiding this comment.
I would say something along the lines of raising an informative error if inferencedata is passed
| @@ -139,6 +139,16 @@ def test_from_pymc_predictions(self, data, eight_schools_params): | |||
| assert ivalues.shape[0] == 1 # one chain in predictions | |||
| assert np.all(np.isclose(ivalues[0], values)) | |||
|
|
|||
| def test_from_pymc_trace_inference_data(self): | |||
| # check if error is raised successfully after passing InferenceData as trace | |||
There was a problem hiding this comment.
This should be a docstring using """ instead of a comment
| with pm.Model(): | ||
| p = pm.Uniform("p", 0, 1) | ||
| pm.Binomial("w", p=p, n=2, observed=1) | ||
| trace = pm.sample(100, chains=2, return_inferencedata=True) | ||
| assert isinstance(trace, InferenceData) |
There was a problem hiding this comment.
I think we don't need to sample from a model, we can use an example inferencedata or create it with from_dict and then do az.from_pymc3(idata, model=pm.Model()) directly
* enabled type checking for io_pymc3 input * Added test and updated changelog * Fixed test and updated changelog msg
Description
Enabled type checking for
io_pymc3input... Now it raises an error whenever input type is ofInferenceData.Fix #1617
Checklist