Incompatibility with scikit-learn 1.2+: Deprecation of if_delegate_has_method in eli5 library

[e5science]

I've encountered an import error when trying to use the eli5 library with the latest version of scikit-learn (1.2+). The error message is as follows:
ImportError: cannot import name 'if_delegate_has_method' from 'sklearn.utils.metaestimators' (path_to_sklearn\utils\metaestimators.py)

The if_delegate_has_method decorator has been deprecated and removed from scikit-learn in version 1.2, and it has been replaced by the available_if decorator. This issue arises when attempting to use eli5 for permutation importance with a trained model.

Steps to Reproduce:

Install the latest versions of scikit-learn and eli5.
Attempt to import eli5 and use PermutationImportance.
Observe the ImportError.

Proposed Solution:

Update the import statements in the eli5 library to use the new available_if decorator from scikit-learn's metaestimators module.

Environment:

Python version: 3.x
scikit-learn version: 1.2+
eli5 version: latest

[DrTomWilliams]

I have a fix for this by manually defining the depreciated function within permutation_importance.py and redirecting some of the eli5 imports to the official sklearn repos.

In "PYTHON_INSTALL_PATH/site-packages/eli5/sklearn/permutation_importance.py" (replace PYTHON_INSTALL_PATH" with where you have python installed; for me it is "~./local/lib/python3.12").

The ammended code is:

# -*- coding: utf-8 -*-
from functools import partial
from typing import List

import numpy as np
from sklearn.model_selection import check_cv

from sklearn.utils import check_array, check_random_state
from sklearn.base import (
    BaseEstimator,
    MetaEstimatorMixin,
    clone,
    is_classifier
)
from sklearn.metrics import check_scoring

from eli5.permutation_importance import get_score_importances
from eli5.sklearn.utils import pandas_available

if pandas_available:
    import pandas as pd






# IMPORTED PREVIOUS VERSION (1.1?) OF SKLEARN.UTILS.METAESTIMATORS.PY
"""Utilities for meta-estimators"""
# Author: Joel Nothman
#         Andreas Mueller
# License: BSD
from typing import List, Any
from types import MethodType
import warnings
from functools import wraps

from abc import ABCMeta, abstractmethod
from operator import attrgetter
from functools import update_wrapper
import numpy as np
from contextlib import suppress

#from ..utils import _safe_indexing
from sklearn.utils import _safe_indexing
from sklearn.utils._tags import _safe_tags
from sklearn.base import BaseEstimator

__all__ = ["available_if", "if_delegate_has_method"]


class _BaseComposition(BaseEstimator, metaclass=ABCMeta):
    """Handles parameter management for classifiers composed of named estimators."""

    steps: List[Any]

    @abstractmethod
    def __init__(self):
        pass

    def _get_params(self, attr, deep=True):
        out = super().get_params(deep=deep)
        if not deep:
            return out

        estimators = getattr(self, attr)
        try:
            out.update(estimators)
        except (TypeError, ValueError):
            # Ignore TypeError for cases where estimators is not a list of
            # (name, estimator) and ignore ValueError when the list is not
            # formatted correctly. This is to prevent errors when calling
            # `set_params`. `BaseEstimator.set_params` calls `get_params` which
            # can error for invalid values for `estimators`.
            return out

        for name, estimator in estimators:
            if hasattr(estimator, "get_params"):
                for key, value in estimator.get_params(deep=True).items():
                    out["%s__%s" % (name, key)] = value
        return out

    def _set_params(self, attr, **params):
        # Ensure strict ordering of parameter setting:
        # 1. All steps
        if attr in params:
            setattr(self, attr, params.pop(attr))
        # 2. Replace items with estimators in params
        items = getattr(self, attr)
        if isinstance(items, list) and items:
            # Get item names used to identify valid names in params
            # `zip` raises a TypeError when `items` does not contains
            # elements of length 2
            with suppress(TypeError):
                item_names, _ = zip(*items)
                for name in list(params.keys()):
                    if "__" not in name and name in item_names:
                        self._replace_estimator(attr, name, params.pop(name))

        # 3. Step parameters and other initialisation arguments
        super().set_params(**params)
        return self

    def _replace_estimator(self, attr, name, new_val):
        # assumes `name` is a valid estimator name
        new_estimators = list(getattr(self, attr))
        for i, (estimator_name, _) in enumerate(new_estimators):
            if estimator_name == name:
                new_estimators[i] = (name, new_val)
                break
        setattr(self, attr, new_estimators)

    def _validate_names(self, names):
        if len(set(names)) != len(names):
            raise ValueError("Names provided are not unique: {0!r}".format(list(names)))
        invalid_names = set(names).intersection(self.get_params(deep=False))
        if invalid_names:
            raise ValueError(
                "Estimator names conflict with constructor arguments: {0!r}".format(
                    sorted(invalid_names)
                )
            )
        invalid_names = [name for name in names if "__" in name]
        if invalid_names:
            raise ValueError(
                "Estimator names must not contain __: got {0!r}".format(invalid_names)
            )


class _AvailableIfDescriptor:
    """Implements a conditional property using the descriptor protocol.

    Using this class to create a decorator will raise an ``AttributeError``
    if check(self) returns a falsey value. Note that if check raises an error
    this will also result in hasattr returning false.

    See https://docs.python.org/3/howto/descriptor.html for an explanation of
    descriptors.
    """

    def __init__(self, fn, check, attribute_name):
        self.fn = fn
        self.check = check
        self.attribute_name = attribute_name

        # update the docstring of the descriptor
        update_wrapper(self, fn)

    def __get__(self, obj, owner=None):
        attr_err = AttributeError(
            f"This {repr(owner.__name__)} has no attribute {repr(self.attribute_name)}"
        )
        if obj is not None:
            # delegate only on instances, not the classes.
            # this is to allow access to the docstrings.
            if not self.check(obj):
                raise attr_err
            out = MethodType(self.fn, obj)

        else:
            # This makes it possible to use the decorated method as an unbound method,
            # for instance when monkeypatching.
            @wraps(self.fn)
            def out(*args, **kwargs):
                if not self.check(args[0]):
                    raise attr_err
                return self.fn(*args, **kwargs)

        return out


def available_if(check):
    """An attribute that is available only if check returns a truthy value

    Parameters
    ----------
    check : callable
        When passed the object with the decorated method, this should return
        a truthy value if the attribute is available, and either return False
        or raise an AttributeError if not available.

    Examples
    --------
    >>> from sklearn.utils.metaestimators import available_if
    >>> class HelloIfEven:
    ...    def __init__(self, x):
    ...        self.x = x
    ...
    ...    def _x_is_even(self):
    ...        return self.x % 2 == 0
    ...
    ...    @available_if(_x_is_even)
    ...    def say_hello(self):
    ...        print("Hello")
    ...
    >>> obj = HelloIfEven(1)
    >>> hasattr(obj, "say_hello")
    False
    >>> obj.x = 2
    >>> hasattr(obj, "say_hello")
    True
    >>> obj.say_hello()
    Hello
    """
    return lambda fn: _AvailableIfDescriptor(fn, check, attribute_name=fn.__name__)


# TODO(1.3) remove
class _IffHasAttrDescriptor(_AvailableIfDescriptor):
    """Implements a conditional property using the descriptor protocol.

    Using this class to create a decorator will raise an ``AttributeError``
    if none of the delegates (specified in ``delegate_names``) is an attribute
    of the base object or the first found delegate does not have an attribute
    ``attribute_name``.

    This allows ducktyping of the decorated method based on
    ``delegate.attribute_name``. Here ``delegate`` is the first item in
    ``delegate_names`` for which ``hasattr(object, delegate) is True``.

    See https://docs.python.org/3/howto/descriptor.html for an explanation of
    descriptors.
    """

    def __init__(self, fn, delegate_names, attribute_name):
        super().__init__(fn, self._check, attribute_name)
        self.delegate_names = delegate_names

    def _check(self, obj):
        warnings.warn(
            "if_delegate_has_method was deprecated in version 1.1 and will be "
            "removed in version 1.3. Use available_if instead.",
            FutureWarning,
        )

        delegate = None
        for delegate_name in self.delegate_names:
            try:
                delegate = attrgetter(delegate_name)(obj)
                break
            except AttributeError:
                continue

        if delegate is None:
            return False
        # raise original AttributeError
        getattr(delegate, self.attribute_name)

        return True


# TODO(1.3) remove
def if_delegate_has_method(delegate):
    """Create a decorator for methods that are delegated to a sub-estimator

    This enables ducktyping by hasattr returning True according to the
    sub-estimator.

    .. deprecated:: 1.3
        `if_delegate_has_method` is deprecated in version 1.1 and will be removed in
        version 1.3. Use `available_if` instead.

    Parameters
    ----------
    delegate : str, list of str or tuple of str
        Name of the sub-estimator that can be accessed as an attribute of the
        base object. If a list or a tuple of names are provided, the first
        sub-estimator that is an attribute of the base object will be used.

    """
    if isinstance(delegate, list):
        delegate = tuple(delegate)
    if not isinstance(delegate, tuple):
        delegate = (delegate,)

    return lambda fn: _IffHasAttrDescriptor(fn, delegate, attribute_name=fn.__name__)


def _safe_split(estimator, X, y, indices, train_indices=None):
    """Create subset of dataset and properly handle kernels.

    Slice X, y according to indices for cross-validation, but take care of
    precomputed kernel-matrices or pairwise affinities / distances.

    If ``estimator._pairwise is True``, X needs to be square and
    we slice rows and columns. If ``train_indices`` is not None,
    we slice rows using ``indices`` (assumed the test set) and columns
    using ``train_indices``, indicating the training set.

    Labels y will always be indexed only along the first axis.

    Parameters
    ----------
    estimator : object
        Estimator to determine whether we should slice only rows or rows and
        columns.

    X : array-like, sparse matrix or iterable
        Data to be indexed. If ``estimator._pairwise is True``,
        this needs to be a square array-like or sparse matrix.

    y : array-like, sparse matrix or iterable
        Targets to be indexed.

    indices : array of int
        Rows to select from X and y.
        If ``estimator._pairwise is True`` and ``train_indices is None``
        then ``indices`` will also be used to slice columns.

    train_indices : array of int or None, default=None
        If ``estimator._pairwise is True`` and ``train_indices is not None``,
        then ``train_indices`` will be use to slice the columns of X.

    Returns
    -------
    X_subset : array-like, sparse matrix or list
        Indexed data.

    y_subset : array-like, sparse matrix or list
        Indexed targets.

    """
    if _safe_tags(estimator, key="pairwise"):
        if not hasattr(X, "shape"):
            raise ValueError(
                "Precomputed kernels or affinity matrices have "
                "to be passed as arrays or sparse matrices."
            )
        # X is a precomputed square kernel matrix
        if X.shape[0] != X.shape[1]:
            raise ValueError("X should be a square kernel matrix")
        if train_indices is None:
            X_subset = X[np.ix_(indices, indices)]
        else:
            X_subset = X[np.ix_(indices, train_indices)]
    else:
        X_subset = _safe_indexing(X, indices)

    if y is not None:
        y_subset = _safe_indexing(y, indices)
    else:
        y_subset = None

    return X_subset, y_subset








CAVEATS_CV_NONE = """
Feature importances are computed on the same data as used for training, 
i.e. feature importances don't reflect importance of features for 
generalization.
"""

CAVEATS_CV = """
Feature importances are not computed for the final estimator; 
they are computed for a sequence of estimators trained and evaluated 
on train/test splits. So they tell you about importances of features 
for generalization, but not feature importances of a particular trained model.
"""

CAVEATS_PREFIT = """
If feature importances are computed on the same data as used for training, 
they don't reflect importance of features for generalization. Use a held-out
dataset if you want generalization feature importances.
"""


class PermutationImportance(BaseEstimator, MetaEstimatorMixin):
    """Meta-estimator which computes ``feature_importances_`` attribute
    based on permutation importance (also known as mean score decrease).

    :class:`~PermutationImportance` instance can be used instead of
    its wrapped estimator, as it exposes all estimator's common methods like
    ``predict``.

    There are 3 main modes of operation:

    1. cv="prefit" (pre-fit estimator is passed). You can call
       PermutationImportance.fit either with training data, or
       with a held-out dataset (in the latter case ``feature_importances_``
       would be importances of features for generalization). After the fitting
       ``feature_importances_`` attribute becomes available, but the estimator
       itself is not fit again. When cv="prefit",
       :meth:`~PermutationImportance.fit` must be called
       directly, and :class:`~PermutationImportance` cannot be used with
       ``cross_val_score``, ``GridSearchCV`` and similar utilities that clone
       the estimator.
    2. cv=None. In this case :meth:`~PermutationImportance.fit` method fits
       the estimator and computes feature importances on the same data, i.e.
       feature importances don't reflect importance of features for
       generalization.
    3. all other ``cv`` values. :meth:`~PermutationImportance.fit` method
       fits the estimator, but instead of computing feature importances for
       the concrete estimator which is fit, importances are computed for
       a sequence of estimators trained and evaluated on train/test splits
       according to ``cv``, and then averaged. This is more resource-intensive
       (estimators are fit multiple times), and importances are not computed
       for the final estimator, but ``feature_importances_`` show importances
       of features for generalization.

    Mode (1) is most useful for inspecting an existing estimator; modes
    (2) and (3) can be also used for feature selection, e.g. together with
    sklearn's SelectFromModel or RFE.

    Currently :class:`~PermutationImportance` works with dense data.

    Parameters
    ----------
    estimator : object
        The base estimator. This can be both a fitted
        (if ``prefit`` is set to True) or a non-fitted estimator.

    scoring : string, callable or None, default=None
        Scoring function to use for computing feature importances.
        A string with scoring name (see scikit-learn `docs`_) or
        a scorer callable object / function with signature
        ``scorer(estimator, X, y)``.
        If ``None``, the ``score`` method of the estimator is used.

        .. _docs: https://scikit-learn.org/stable/modules/model_evaluation.html#common-cases-predefined-values

    n_iter : int, default 5
        Number of random shuffle iterations. Decrease to improve speed,
        increase to get more precise estimates.

    random_state : integer or numpy.random.RandomState, optional
        random state

    cv : int, cross-validation generator, iterable or "prefit"
        Determines the cross-validation splitting strategy.
        Possible inputs for cv are:

            - None, to disable cross-validation and compute feature importances
              on the same data as used for training.
            - integer, to specify the number of folds.
            - An object to be used as a cross-validation generator.
            - An iterable yielding train/test splits.
            - "prefit" string constant (default).

        If "prefit" is passed, it is assumed that ``estimator`` has been
        fitted already and all data is used for computing feature importances.

    refit : bool
        Whether to fit the estimator on the whole data if cross-validation
        is used (default is True).

    Attributes
    ----------
    feature_importances_ : array
        Feature importances, computed as mean decrease of the score when
        a feature is permuted (i.e. becomes noise).

    feature_importances_std_ : array
        Standard deviations of feature importances.

    results_ : list of arrays
        A list of score decreases for all experiments.

    scores_ : array of float
        A list of base scores for all experiments (with no features permuted).

    estimator_ : an estimator
        The base estimator from which the :class:`~PermutationImportance`
        instance  is built. This is stored only when a non-fitted estimator
        is passed to the :class:`~PermutationImportance`, i.e when ``cv`` is
        not "prefit".

    rng_ : numpy.random.RandomState
        random state
    """
    def __init__(self, estimator, scoring=None, n_iter=5, random_state=None,
                 cv='prefit', refit=True):
        # type: (...) -> None
        if isinstance(cv, str) and cv != "prefit":
            raise ValueError("Invalid cv value: {!r}".format(cv))
        self.refit = refit
        self.estimator = estimator
        self.scoring = scoring
        self.n_iter = n_iter
        self.random_state = random_state
        self.cv = cv
        self.rng_ = check_random_state(random_state)

    def _wrap_scorer(self, base_scorer, pd_columns):
        def pd_scorer(model, X, y):
            X = pd.DataFrame(X, columns=pd_columns)
            return base_scorer(model, X, y)
        return pd_scorer

    def fit(self, X, y, groups=None, **fit_params):
        # type: (...) -> PermutationImportance
        """Compute ``feature_importances_`` attribute and optionally
        fit the base estimator.

        Parameters
        ----------
        X : array-like of shape (n_samples, n_features)
            The training input samples.

        y : array-like, shape (n_samples,)
            The target values (integers that correspond to classes in
            classification, real numbers in regression).

        groups : array-like, with shape (n_samples,), optional
            Group labels for the samples used while splitting the dataset into
            train/test set.

        **fit_params : Other estimator specific parameters

        Returns
        -------
        self : object
            Returns self.
        """
        self.scorer_ = check_scoring(self.estimator, scoring=self.scoring)

        if pandas_available and isinstance(X, pd.DataFrame):
            self.scorer_ = self._wrap_scorer(self.scorer_, X.columns)

        if self.cv != "prefit" and self.refit:
            self.estimator_ = clone(self.estimator)
            self.estimator_.fit(X, y, **fit_params)

        X = check_array(X, force_all_finite='allow-nan')

        if self.cv not in (None, "prefit"):
            si = self._cv_scores_importances(X, y, groups=groups, **fit_params)
        else:
            si = self._non_cv_scores_importances(X, y)
        scores, results = si
        self.scores_ = np.array(scores)
        self.results_ = results
        self.feature_importances_ = np.mean(results, axis=0)
        self.feature_importances_std_ = np.std(results, axis=0)
        return self

    def _cv_scores_importances(self, X, y, groups=None, **fit_params):
        assert self.cv is not None
        cv = check_cv(self.cv, y, classifier=is_classifier(self.estimator))
        feature_importances = []  # type: List
        base_scores = []  # type: List[float]
        weights = fit_params.pop('sample_weight', None)
        fold_fit_params = fit_params.copy()
        for train, test in cv.split(X, y, groups):
            if weights is not None:
                fold_fit_params['sample_weight'] = weights[train]
            est = clone(self.estimator).fit(X[train], y[train], **fold_fit_params)
            score_func = partial(self.scorer_, est)
            _base_score, _importances = self._get_score_importances(
                score_func, X[test], y[test])
            base_scores.extend([_base_score] * len(_importances))
            feature_importances.extend(_importances)
        return base_scores, feature_importances

    def _non_cv_scores_importances(self, X, y):
        score_func = partial(self.scorer_, self.wrapped_estimator_)
        base_score, importances = self._get_score_importances(score_func, X, y)
        return [base_score] * len(importances), importances

    def _get_score_importances(self, score_func, X, y):
        return get_score_importances(score_func, X, y, n_iter=self.n_iter,
                                     random_state=self.rng_)


    @property
    def caveats_(self):
        # type: () -> str
        if self.cv == 'prefit':
            return CAVEATS_PREFIT
        elif self.cv is None:
            return CAVEATS_CV_NONE
        return CAVEATS_CV

    # ============= Exposed methods of a wrapped estimator:

    @if_delegate_has_method(delegate='wrapped_estimator_')
    def score(self, X, y=None, *args, **kwargs):
        return self.wrapped_estimator_.score(X, y, *args, **kwargs)

    @if_delegate_has_method(delegate='wrapped_estimator_')
    def predict(self, X):
        return self.wrapped_estimator_.predict(X)

    @if_delegate_has_method(delegate='wrapped_estimator_')
    def predict_proba(self, X):
        return self.wrapped_estimator_.predict_proba(X)

    @if_delegate_has_method(delegate='wrapped_estimator_')
    def predict_log_proba(self, X):
        return self.wrapped_estimator_.predict_log_proba(X)

    @if_delegate_has_method(delegate='wrapped_estimator_')
    def decision_function(self, X):
        return self.wrapped_estimator_.decision_function(X)

    @property
    def wrapped_estimator_(self):
        if self.cv == "prefit" or not self.refit:
            return self.estimator
        return self.estimator_

    @property
    def _estimator_type(self):
        return self.estimator._estimator_type

    @property
    def classes_(self):
        return self.wrapped_estimator_.classes_

[lopuhin]

hi, this is fixed in master branch of https://github.com/eli5-org/eli5 and we'll do a PyPI release soon

[lopuhin]

this is fixed in the latest release on PyPI: 0.14.0

[github-actions]

This issue has been marked stale and will be closed in 7 days unless updated.