Merge pull request #281 from mdekstrand/feature/split-implicit

Split implicit wrappers into new package

.github/workflows/test-suite.yml

@@ -53,7 +53,7 @@ jobs:
           if [ $RUNNER = Linux -a $PYVER = 3.8 ]; then
             echo "Using most extras"
             # we'll test tensorflow in a separate run
-            e_opt="-e implicit,sklearn"
+            e_opt="-e sklearn"
           fi
           bash lkbuild/lock-for-ci.sh -v $PYVER -b $BLAS_IMPL $e_opt
         env:

docs/addons.rst

@@ -7,4 +7,5 @@ to implementations in other packages.
 .. toctree::
     :caption: External Algorithms
 
+    Implicit wrappers <https://lkpy.lenskit.org/projects/lenskit-implicit>
     Poisson factorization <https://lkpy.lenskit.org/projects/lenskit-hpf/>

docs/algorithms.rst

@@ -55,8 +55,3 @@ Add-On Packages
 
 See `add-on algorithms <addons.rst>`_ for additional algorithm families and bridges to other
 packages.
-
-.. autosummary::
-
-    implicit.BPR
-    implicit.ALS

lenskit/algorithms/implicit.py

@@ -1,125 +1 @@
-import logging
-import inspect
-import pandas as pd
-import numpy as np
-
-from ..data import sparse_ratings
-from . import Recommender, Predictor
-
-_logger = logging.getLogger(__name__)
-
-
-class BaseRec(Recommender, Predictor):
-    """
-    Base class for Implicit-backed recommenders.
-
-    Args:
-        delegate(implicit.RecommenderBase):
-            The delegate algorithm.
-
-    Attributes:
-        delegate(implicit.RecommenderBase):
-            The :py:mod:`implicit` delegate algorithm.
-        matrix_(scipy.sparse.csr_matrix):
-            The user-item rating matrix.
-        user_index_(pandas.Index):
-            The user index.
-        item_index_(pandas.Index):
-            The item index.
-    """
-
-    def __init__(self, delegate):
-        self.delegate = delegate
-
-    def fit(self, ratings, **kwargs):
-        matrix, users, items = sparse_ratings(ratings, scipy=True)
-        iur = matrix.T.tocsr()
-
-        _logger.info('training %s on %s matrix (%d nnz)', self.delegate, iur.shape, iur.nnz)
-
-        self.delegate.fit(iur)
-
-        self.matrix_ = matrix
-        self.user_index_ = users
-        self.item_index_ = items
-
-        return self
-
-    def recommend(self, user, n=None, candidates=None, ratings=None):
-        try:
-            uid = self.user_index_.get_loc(user)
-        except KeyError:
-            return pd.DataFrame({'item': []})
-
-        if candidates is None:
-            i_n = n if n is not None else len(self.item_index_)
-            recs = self.delegate.recommend(uid, self.matrix_, N=i_n)
-        else:
-            cands = self.item_index_.get_indexer(candidates)
-            cands = cands[cands >= 0]
-            recs = self.delegate.rank_items(uid, self.matrix_, cands)
-
-        if n is not None:
-            recs = recs[:n]
-        rec_df = pd.DataFrame.from_records(recs, columns=['item_pos', 'score'])
-        rec_df['item'] = self.item_index_[rec_df.item_pos]
-        return rec_df.loc[:, ['item', 'score']]
-
-    def predict_for_user(self, user, items, ratings=None):
-        try:
-            uid = self.user_index_.get_loc(user)
-        except KeyError:
-            return pd.Series(np.nan, index=items)
-
-        iids = self.item_index_.get_indexer(items)
-        iids = iids[iids >= 0]
-
-        ifs = self.delegate.item_factors[iids]
-        uf = self.delegate._user_factor(uid, None, False)
-        scores = ifs.dot(uf)
-        scores = pd.Series(scores, index=self.item_index_[iids])
-        return scores.reindex(items)
-
-    def __getattr__(self, name):
-        if 'delegate' not in self.__dict__:
-            raise AttributeError()
-        dd = self.delegate.__dict__
-        if name in dd:
-            return dd[name]
-        else:
-            raise AttributeError()
-
-    def get_params(self, deep=True):
-        dd = self.delegate.__dict__
-        sig = inspect.signature(self.delegate.__class__)
-        names = list(sig.parameters.keys())
-        return dict([(k, dd.get(k)) for k in names])
-
-    def __str__(self):
-        return 'Implicit({})'.format(self.delegate)
-
-
-class ALS(BaseRec):
-    """
-    LensKit interface to :py:mod:`implicit.als`.
-    """
-    def __init__(self, *args, **kwargs):
-        """
-        Construct an ALS recommender.  The arguments are passed as-is to
-        :py:class:`implicit.als.AlternatingLeastSquares`.
-        """
-        from implicit.als import AlternatingLeastSquares
-        super().__init__(AlternatingLeastSquares(*args, **kwargs))
-
-
-class BPR(BaseRec):
-    """
-    LensKit interface to :py:mod:`implicit.bpr`.
-    """
-    def __init__(self, *args, **kwargs):
-        """
-        Construct an ALS recommender.  The arguments are passed as-is to
-        :py:class:`implicit.als.BayesianPersonalizedRanking`.
-        """
-        from implicit.bpr import BayesianPersonalizedRanking
-        super().__init__(BayesianPersonalizedRanking(*args, **kwargs))
+from lenskit_implicit import BaseRec, ALS, BPR  # noqa: F401

pyproject.toml

@@ -68,10 +68,10 @@ demo = [
     "nbval >= 0.9",
     "matplotlib ~= 3.4",
 ]
-hpf = ["lenskit-hpf"]
-implicit = ["implicit"]
 sklearn = ["scikit-learn >= 0.22"]
 tf = ["tensorflow >=2.1,<2.6"]
+hpf = ["lenskit-hpf"]
+implicit = ["lenskit-implicit"]
 
 [tool.flit.sdist]
 exclude = [