This is major release that modifies the MERF API in a backward-incompatible way. There are a number of things added to this release. The highlights:
- Added ability to add any fixed effects learner, e.g. gradient boosting, deep learning
- Added ability to add validation set and track validation loss during training process
- Updated code to work with pandas 1.0+
- Added a plotting utility to look at results of training
- Added documentation built using Sphinx
- Added more unit tests for pickling, typing, plotting, etc.
- Small bug fixes
- Moved CI over to GitHub Actions
- Published documentation using GitHub pages
- Moved over to Orbyter 2.0
- Moved to Makefile for builds, CI, linting