Skip to content

sklearn/metadata-routing

Rule metadata-routing
Pack scikit-learn
Severity error
Fix unsafe, with xpd check --fix --unsafe-fixes
Library versions every version

These forms work only with metadata routing on, or only with it off, so the call raises RuntimeError or ValueError here. Turn routing on with sklearn.set_config(enable_metadata_routing=True), or use the form for the current setting.

model = LogisticRegression().set_fit_request(sample_weight=True)
sklearn.set_config(enable_metadata_routing=True)
model = LogisticRegression().set_fit_request(sample_weight=True).set_score_request(sample_weight=True)
cross_validate(model, X, y, params={"sample_weight": w})

The rule reports forms that raise because metadata routing is off, such as set_fit_request and Pipeline(transform_input=...). It also reports forms that raise because routing is on, such as groups=, step__sample_weight= and params= with no request in the file.

xpd check --fix --unsafe-fixes applies the fixes of this rule. Check the result, because an unsafe fix can change what the code does.

The unsafe fix writes params={"groups": groups} for groups= when routing is on, because the old form raises. The other forms have no fix.

xpd applies the replacement with --fix --unsafe-fixes.

The rule counts a file as routed when any call in it passes enable_metadata_routing with a value other than False. A function or a script that never calls a set_*_request method has asked for nothing.

Routing Form Raises
off set_fit_request and the other request setters RuntimeError
off Pipeline(transform_input=...) ValueError at fit
on groups= on a cross-validation function ValueError
on a step__sample_weight keyword on fit TypeError
on params= when no estimator asked for the metadata UnsetMetadataPassedError

On scikit-learn 1.9.1 a weighted cross_validate with routing on also needs set_score_request, or it raises for the scorer. Recall was 34 of 34.

xpd prints this message with the issue. A name that starts with $ stands for code from your file.

$M works only when metadata routing is on, so it raises RuntimeError here. Call sklearn.set_config(enable_metadata_routing=True) before you build the estimator.

Browse the other rules of the scikit-learn pack.