Invert order of CV so it runs the largest first

[fraimondo]

When the final model is expected, the CV iterator ads one more split that trains on all data.

This PR inverts the order, so the first split is the final model.

This has no impact on the results, but has an impact on compute time when masively parallelised (e.g. using htcondor).

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 89.93%. Comparing base (d0fcf31) to head (ffca079).
Report is 6 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #275   +/-   ##
=======================================
  Coverage   89.93%   89.93%           
=======================================
  Files          56       56           
  Lines        2493     2493           
  Branches      443      443           
=======================================
  Hits         2242     2242           
  Misses        165      165           
  Partials       86       86           

Flag	Coverage Δ
docs	100.00% <ø> (ø)
julearn	89.92% <100.00%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
julearn/api.py	92.89% <100.00%> (ø)
julearn/model_selection/final_model_cv.py	100.00% <100.00%> (ø)

[synchon]

Maybe have the version changes of pyproject.toml in a separate PR to keep things traceable?

[github-actions]

PR Preview Action v1.4.8
Preview removed because the pull request was closed.
2024-10-17 13:59 UTC