search

Aarhus-Psychiatry-Research / timeseriesflattener

Converting irregularly spaced time series, such as eletronic health records, into dataframes for tabular classification.
https://Aarhus-Psychiatry-Research.github.io/timeseriesflattener
MIT License
19 stars 2 forks source link

Feat/490/make_user_facing_specs_and_Flattener_easily_importable_and_put_non-userfacing_into_private_scope #493

Closed HLasse closed 8 months ago

HLasse commented 8 months ago

Left aggregators out - seems more intuitive to do from timeseriesflattener.aggregators import ...

Unsure whether to put the PredictorGroupSpec from from_legacy.py into top level. Ideally, we don't really want people to use (except perhaps our group for ease of transitioning). Didn't put it for now

codspeed-hq[bot] commented 8 months ago

CodSpeed Performance Report

Merging #493 will not alter performance

Comparing feat/490/make_user_facing_specs_and_Flattener_easily_importable_and_put_non-userfacing_into_private_scope (94752f4) with main (7e4f0d7)

Summary

✅ 7 untouched benchmarks