Summary of this 2023-03 release:
- It is no longer necessary to specify categorical variables yourself when calling Cobra's preprocessing, an automatic search of categorical variables can be done.
- It is not longer obligatory to specify a row ID column for the inputted basetable.
- We now automatically drop columns that only contain missing values, which in the past caused Cobra throwing exotic errors when preprocessing the basetable.
- We prevent fit_transform() of modifying the training data dataframe under some circumstances.
- As a short-term fix for pandas's 2.0 release, which breaks a small portion of Cobra's code, we impose that Cobra is installed with pandas < 2.0.0.
List of issues that were behind this: see the closed 2023-03 release milestone.
We welcome the following new contributors!
- @ZlaTanskY made their first contribution in #131
- @patrickleonardy made their first contribution in #142
Full details (autogenerated by Github) of the associated issues and pull requests:
- feat: added files to encourage PEP8 by @ZlaTanskY in #131
- Fix/mutable train data in fit transform by @ZlaTanskY in #140
- Added drop of columns containing only NANs by @patrickleonardy in #142
- Defaults id_column to None for PIGs & tests by @patrickleonardy in #141
- issue 137/PR 145 automatically search for categorical variables by @patrickleonardy in #145
- 137 automatically search for categorical variables by @patrickleonardy in #147
- #151 add turorial while importing by @patrickleonardy in #152
- Short term pandas 2.0 support by @sandervh14 in #164
- Merging 2023-03 development branch to master for 2023-03 release (1.1.1) by @sandervh14 in #162
Full Changelog: v1.1.0...v1.1.1
Contributors
sandervh14, ZlaTanskY, and patrickleonardy