All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog,
and this project is versionned in the YY.MM format corresponding to the month in which the model was first used in production.
- Created the
predictsignauxfaiblespython package - Created
SFDatasetto help query data from MongoDB - Created
OversampledSFDatasetto oversample firms that are in default - Created projet config based on environment.
- Created
is_randomdecorator to help with reproducibility - Created
make_sf_train_test_splitsto help us perform unbiased cros-validation - Created modular pipelines of
Preprocessorsto perform preprocessing tasks - Created ML models configuration files written in python
- Created a CLI that consumes and runs models based on their conf files
- Created basic unit tests
- Created and enforced data science development workflow
- Nothing, it's our first release 😄 🎉
- Nothing, it's our first release 😄 🎉
- Nothing, it's our first release 😄 🎉
- Created documentation using Sphinx
- Created synthetic data generation capabilities
- Created explainability module
- Created a CLI to easily parameterize model training runs
- Created model config files
- Created "redressements experts" capacities starting with a URSSAF-based rule
- Created evaluation module
- fill missing fields after fetching data
- merging redundant operations on data.columns
- force siren and sirets to be strings + pad them to 9 characters
- downgrade jedi version to fix auto-complete bug with ipython.