-
-
Notifications
You must be signed in to change notification settings - Fork 26.9k
MNT Make binary display method parameters' order consistent #30717
Copy link
Copy link
Closed
Labels
good first issueEasy with clear instructions to resolveEasy with clear instructions to resolvemodule:model_selection
Description
This came up while working on #30399 . These are all classes inheriting the _BinaryClassifierCurveDisplayMixin.
RocCurveDisplayandPrecisionRecallDisplayare pretty consistent, we would just need to change wherepos_labelis. No strong preference to where it should be.DetCurveDisplaydoes not have the chance level line,drop_intermediateanddepsine. Chance line is added in ENH/FIX add drop_intermediate to DET curve and add threshold at infinity #29151 (we should ensure order is consistent in that PR). Note there is discussion of addingdrop_intermediatein that PR as wellCalibrationDisplay- is a bit different from the rest, e.g., there is a reference line (perfect calibration) and not a chance line. We could moveaxup though, to be consistent with the other displays.
Table of parameters
| CalibrationDisplay | DetCurveDisplay | RocCurveDisplay | PrecisionRecallDisplay | |
|---|---|---|---|---|
| plot | ax name ref_line kwargs |
ax name kwargs |
ax name plot_chance_level chance_level_kw despine kwargs |
ax name plot_chance_level chance_level_kw despine kwargs |
| from_estimator | estimator X y n_bins strategy pos_label name ref_line ax kwargs |
estimator X y sample_weight response_method pos_label name ax kwargs |
estimator X y sample_weight drop_intermediate response_method pos_label name ax plot_chance_level chance_level_kw despine kwargs |
estimator X y sample_weight pos_label drop_intermediate response_method name ax plot_chance_level chance_level_kw despine kwargs |
| from_predictions | y_true y_prob n_bins strategy pos_label name ref_line ax kwargs |
y_true y_pred sample_weight pos_label name ax kwargs |
y_true y_pred sample_weight drop_intermediate pos_label name ax plot_chance_level chance_level_kw despine kwargs |
y_true y_pred sample_weight pos_label drop_intermediate name ax plot_chance_level chance_level_kw despine kwargs |
Discussed with @DeaMariaLeon @glemaitre
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
good first issueEasy with clear instructions to resolveEasy with clear instructions to resolvemodule:model_selection