Skip to content

category_encoders/deprecated-in-category-encoders

Rule deprecated-in-category-encoders
Pack Category Encoders
Severity warning
Fix safe, with xpd check --fix
Library versions Category Encoders 2.6 and later

These names and arguments warn on category_encoders and will be removed. Use the replacement that the message names, such as get_feature_names_out().

names = encoder.get_feature_names()
hasher = ce.HashingEncoder(cols=["city"], max_sample=1000, n_components=64)
names = encoder.get_feature_names_out()
hasher = ce.HashingEncoder(cols=["city"], n_components=64)

The rule reports encoder.get_feature_names() and HashingEncoder(max_sample=...).

xpd check --fix applies the fix of this rule.

The severity is warning for get_feature_names, and info for max_sample. The fix is safe.

get_feature_names raised FutureWarning: `get_feature_names` is deprecated in all of sklearn. Use `get_feature_names_out` instead. Both methods returned the same NumPy array, so the fix is safe. The method get_feature_names_out was added in 2.6, so the document carries since: "2.6".

scikit-learn removed its own get_feature_names, and removed-in-sklearn reports every call in a file that imports sklearn. deprecated-in-category-encoders therefore runs only in a file that doesn't import sklearn. It needs the receiver built from a class whose name ends in Encoder or Wrapper, in the same expression or in an assignment of the same function.

The changelog of 2.6.4 says of max_sample that "it has no use anymore". HashingEncoder still takes the keyword and raises no warning, so the document is info. With max_sample=10 and without it the encoded frame was equal, with one process and with two, so the fix that drops the keyword is safe.

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

get_feature_names is deprecated in category_encoders and raises FutureWarning. Call $E.get_feature_names_out(), which returns the same array.

Browse the other rules of the Category Encoders pack.