Add more cross-validation options
Open
- Dominant language
- Jupyter Notebook
- Stars
- 201
- Forks
- 62
- PR merge metrics
- No merged PRs in 30d
Description
We do temporal. Add capability for
- standard CV (i.e. randomly selected)
- spatial ( #369 )
- spatio-temporal
- unit (e.g. train on all officers but SWAT team, test on SWAT team)
A lot of the pipeline is pretty hardcoded for temporal validation, some work will have to be done to figure out what changes to the results schema we need to properly store this data.
Contributor guide
Research direction
Start by tracing the existing temporal-validation pipeline and results schema. Determine how standard, spatial, spatio-temporal, and unit validation should be represented, then verify that each mode's results can be stored without relying on temporal-only assumptions.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- data, machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 30/100