Real-world uses

Created for this library

1. 1.

   A research team uses a stratified split on labels so each split has the same proportion of positive examples.

2. 2.

   An ML platform team uses time-based splits for time-series problems so evaluation reflects future-like conditions.

3. 3.

   A risk modeling team uses an out-of-time split so the production model is evaluated on more recent data than it was trained on.