Currently we only have 4 training examples, 1 val example, and 0 test. Let's make a larger meds dataset, run meds-polars on it and copy the output into the tests/test_data folder. Ideally we should have around 128 patients with multiple tasks defined (mortality and los).
Currently we only have 4 training examples, 1 val example, and 0 test. Let's make a larger meds dataset, run meds-polars on it and copy the output into the
tests/test_data
folder. Ideally we should have around 128 patients with multiple tasks defined (mortality and los).We can just duplicate rows from this branch of meds_polars for now: https://github.com/mmcdermott/MEDS_polars_functions/blob/dfd0fa7fe9d844121d0a9dce5f71d68e6340b9df/tests/test_preprocessing.py#L131C19-L131C27