This work accompanies the corresponding Scratch repository: https://github.com/NetherlandsForensicInstitute/scratch. Within this specific repository, the LR module - responsible for calculating the appropriate statistical data - is provided.
From the lrmodule python module, several public API methods are exposed:
- ... (TODO)
- Install all dependencies using
pdm sync -G dev(to install dev dependencies as well) - Run checks with
pdm run checkorpdm check-qualityto automatically fix the things as well - Run tests with
pdm run test - To run everything
pdm run all(no auto fixes) or with fixes:pdm run all-fix
All typing, linting and formatting configuration was taken from the Scratch repository for seamless integration.
- Update hyperparameters, experiment setup and data path as needed in
validation.yaml; - Run validation experiments as
pdm run lir validation.yaml; - Inspect the results in the output folder;
- When satisfied, update the stored model (TODO).