Change Log

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

[0.1.4] - 2020-12-24

Supports numpyro 0.4.0 and the jax updates therein! Also adds a check for null data in fit/predict and raises an exception that makes the problem obvious.

[0.1.3] - 2020-07-27

0.1.2 came a day too soon! Numpyro created a new release (0.3.0) with stable dependencies and this release migrates to that version.

[0.1.2] - 2020-07-26

Another bugfix release, no changes to the API. Numpyro’s dependencies have not stablized yet so I am pinning to a specific commit (the latest one at the time I pinned it).

  • Continued Fixes #12: reduce hackyness of the numpyro/jax installation step in CI.

[0.1.1] - 2020-05-16

A bugfix release, no changes to the API. Users with older versions of jax/jaxlib may need to upgrade to jax~=0.1.67 / jaxlib~=0.1.47 as some parts of the jax API have changed. Shabadoo has changed how it imports functions from jax and will not import correctly with older jax versions.

  • Fixes #24: set rng_key for tests that involve randomness.

  • Fixes #12: reduce hackyness of the numpyro/jax installation step in CI.

[0.1.0] - 2020-04-06

  • Fixes #20 - Preserve chain grouping on .fit. Most other changes inservice of this.

  • Alter json export and import to support 2d samples.

  • Change all tests to use 2d samples.

  • Nest samples within a config dict to allow new fields for import/export.

  • Use custom JSONEncoder.

  • Nix pre_from_samples in favor of class attributes.

  • Rename from_samples to from_dict.

  • Make distinct config preprocessor.

[0.0.4] - 2020-03-07

  • Fixes #17: Support an aggfunc option for .predict

  • Fixes #15: Custom errors for better exception catching

  • Add sampler_kwargs to .fit() so that users can tune the sampler.

[0.0.3] - 2020-02-23

  • Fixes #14 - A bug in which the formula printer did not include plus signs.

  • Fixes #11 - Allow users to set the rng seed when randomness is used (.fit, .sample_posterior_predictive). Thanks @favianrahman for the suggestion!

  • Fixes #6 - Add a confidence interval option in the .predict method.

  • Add this changelog!

[0.0.1, 0.0.2, 0.0.2.1] - 2020-02-15

  • Init releases. I made many because there were some problems with the pypi auto release.