alteryx / autonormalize

python library for automated dataset normalization
https://blog.featurelabs.com/automatic-dataset-normalization-for-feature-engineering-in-python/
BSD 3-Clause "New" or "Revised" License
111 stars 16 forks source link

New index columns are not deterministic #24

Closed thehomebrewnerd closed 4 years ago

thehomebrewnerd commented 4 years ago

New index column names are not deterministic as they are created by joining together items in an unsorted list in make_indexes() in normalize.py. This should be updated so that new index names are consistent from run to run. This can be accomplished by sorting the list prior to joining into a new column name.

After making this update, the tests implemented in PR #12 should also be reviewed and updated if needed.

thehomebrewnerd commented 4 years ago

Closed by #25