[WIP] Changes in sklearn wrappers for LDA and LSI models

gensim/sklearn_integration/sklearn_wrapper_gensim_ldamodel.py

@@ -3,35 +3,37 @@
 #
 # Copyright (C) 2011 Radim Rehurek <radimrehurek@seznam.cz>
 # Licensed under the GNU LGPL v2.1 - http://www.gnu.org/licenses/lgpl.html
-#
+
 """
 Scikit learn interface for gensim for easy use of gensim with scikit-learn
 follows on scikit learn API conventions
 """
+
 import numpy as np
+from scipy import sparse
+from sklearn.base import TransformerMixin, BaseEstimator
+from sklearn.exceptions import NotFittedError
 
 from gensim import models
 from gensim import matutils
 from gensim.sklearn_integration import base_sklearn_wrapper
-from scipy import sparse
-from sklearn.base import TransformerMixin, BaseEstimator
 
 
-class SklearnWrapperLdaModel(models.LdaModel, base_sklearn_wrapper.BaseSklearnWrapper, TransformerMixin, BaseEstimator):
+class SklLdaModel(base_sklearn_wrapper.BaseSklearnWrapper, TransformerMixin, BaseEstimator):
     """
     Base LDA module
     """
 
     def __init__(
-            self, corpus=None, num_topics=100, id2word=None,
+            self, num_topics=100, id2word=None,
             chunksize=2000, passes=1, update_every=1,
             alpha='symmetric', eta=None, decay=0.5, offset=1.0,
             eval_every=10, iterations=50, gamma_threshold=0.001,
             minimum_probability=0.01, random_state=None):
         """
         Sklearn wrapper for LDA model. derived class for gensim.model.LdaModel .
         """
-        self.corpus = corpus
+        self.gensim_model = None
         self.num_topics = num_topics
         self.id2word = id2word
         self.chunksize = chunksize
@@ -46,82 +48,66 @@ def __init__(
         self.gamma_threshold = gamma_threshold
         self.minimum_probability = minimum_probability
         self.random_state = random_state
-        # if no fit function is used , then corpus is given in init
-        if self.corpus:
-            models.LdaModel.__init__(
-                self, corpus=self.corpus, num_topics=self.num_topics, id2word=self.id2word,
-                chunksize=self.chunksize, passes=self.passes, update_every=self.update_every,
-                alpha=self.alpha, eta=self.eta, decay=self.decay, offset=self.offset,
-                eval_every=self.eval_every, iterations=self.iterations,
-                gamma_threshold=self.gamma_threshold, minimum_probability=self.minimum_probability,
-                random_state=self.random_state)
 
     def get_params(self, deep=True):
         """
         Returns all parameters as dictionary.
         """
-        return {"corpus": self.corpus, "num_topics": self.num_topics, "id2word": self.id2word,
-                "chunksize": self.chunksize, "passes": self.passes,
-                "update_every": self.update_every, "alpha": self.alpha, "eta": self.eta, "decay": self.decay,
-                "offset": self.offset, "eval_every": self.eval_every, "iterations": self.iterations,
+        return {"num_topics": self.num_topics, "id2word": self.id2word, "chunksize": self.chunksize,
+                "passes": self.passes, "update_every": self.update_every, "alpha": self.alpha, "eta": self.eta,
+                "decay": self.decay, "offset": self.offset, "eval_every": self.eval_every, "iterations": self.iterations,
                 "gamma_threshold": self.gamma_threshold, "minimum_probability": self.minimum_probability,
                 "random_state": self.random_state}
 
     def set_params(self, **parameters):
         """
         Set all parameters.
         """
-        super(SklearnWrapperLdaModel, self).set_params(**parameters)
+        super(SklLdaModel, self).set_params(**parameters)
 
     def fit(self, X, y=None):
         """
-        For fitting corpus into the class object.
-        Calls gensim.model.LdaModel:
-        >>> gensim.models.LdaModel(corpus=corpus, num_topics=num_topics, id2word=id2word, passes=passes, update_every=update_every, alpha=alpha, iterations=iterations, eta=eta, random_state=random_state)
+        Fit the model according to the given training data.
+        Calls gensim.models.LdaModel
         """
         if sparse.issparse(X):
-            self.corpus = matutils.Sparse2Corpus(X)
+            corpus = matutils.Sparse2Corpus(X)
         else:
-            self.corpus = X
+            corpus = X
 
-        models.LdaModel.__init__(
-            self, corpus=self.corpus, num_topics=self.num_topics, id2word=self.id2word,
+        self.gensim_model = models.LdaModel(corpus=corpus, num_topics=self.num_topics, id2word=self.id2word,
             chunksize=self.chunksize, passes=self.passes, update_every=self.update_every,
             alpha=self.alpha, eta=self.eta, decay=self.decay, offset=self.offset,
             eval_every=self.eval_every, iterations=self.iterations,
             gamma_threshold=self.gamma_threshold, minimum_probability=self.minimum_probability,
             random_state=self.random_state)
         return self
 
-    def transform(self, docs, minimum_probability=None):
+    def transform(self, docs):
         """
         Takes as an list of input a documents (documents).
         Returns matrix of topic distribution for the given document bow, where a_ij
         indicates (topic_i, topic_probability_j).
+        The input `docs` should be in BOW format and can be a list of documents like : [ [(4, 1), (7, 1)], [(9, 1), (13, 1)], [(2, 1), (6, 1)] ]
+        or a single document like : [(4, 1), (7, 1)]
         """
+        if self.gensim_model is None:
+            raise NotFittedError("This model has not been fitted yet. Call 'fit' with appropriate arguments before using this method.")
+
         # The input as array of array
         check = lambda x: [x] if isinstance(x[0], tuple) else x
         docs = check(docs)
         X = [[] for _ in range(0, len(docs))]
 
         for k, v in enumerate(docs):
-            doc_topics = self.get_document_topics(v, minimum_probability=minimum_probability)
+            doc_topics = self.gensim_model[v]
             probs_docs = list(map(lambda x: x[1], doc_topics))
             # Everything should be equal in length
             if len(probs_docs) != self.num_topics:
                 probs_docs.extend([1e-12]*(self.num_topics - len(probs_docs)))
             X[k] = probs_docs
         return np.reshape(np.array(X), (len(docs), self.num_topics))
 
-    def get_topic_dist(self, bow, minimum_probability=None, minimum_phi_value=None, per_word_topics=False):
-        """
-        Takes as an input a new document (bow).
-        Returns the topic distribution for the given document bow, as a list of (topic_id, topic_probability) 2-tuples.
-        """
-        return self.get_document_topics(
-            bow, minimum_probability=minimum_probability,
-            minimum_phi_value=minimum_phi_value, per_word_topics=per_word_topics)
-
     def partial_fit(self, X):
         """
         Train model over X.
@@ -134,4 +120,12 @@ def partial_fit(self, X):
         if sparse.issparse(X):
             X = matutils.Sparse2Corpus(X)
 
-        self.update(corpus=X)
+        if self.gensim_model is None:
+            self.gensim_model = models.LdaModel(num_topics=self.num_topics, id2word=self.id2word,
+                chunksize=self.chunksize, passes=self.passes, update_every=self.update_every,
+                alpha=self.alpha, eta=self.eta, decay=self.decay, offset=self.offset,
+                eval_every=self.eval_every, iterations=self.iterations, gamma_threshold=self.gamma_threshold,
+                minimum_probability=self.minimum_probability, random_state=self.random_state)
+
+        self.gensim_model.update(corpus=X)
+        return self

gensim/sklearn_integration/sklearn_wrapper_gensim_lsimodel.py

@@ -3,31 +3,33 @@
 #
 # Copyright (C) 2011 Radim Rehurek <radimrehurek@seznam.cz>
 # Licensed under the GNU LGPL v2.1 - http://www.gnu.org/licenses/lgpl.html
-#
+
 """
 Scikit learn interface for gensim for easy use of gensim with scikit-learn
 Follows scikit-learn API conventions
 """
+
 import numpy as np
+from scipy import sparse
+from sklearn.base import TransformerMixin, BaseEstimator
+from sklearn.exceptions import NotFittedError
 
 from gensim import models
 from gensim import matutils
 from gensim.sklearn_integration import base_sklearn_wrapper
-from scipy import sparse
-from sklearn.base import TransformerMixin, BaseEstimator
 
 
-class SklearnWrapperLsiModel(models.LsiModel, base_sklearn_wrapper.BaseSklearnWrapper, TransformerMixin, BaseEstimator):
+class SklLsiModel(base_sklearn_wrapper.BaseSklearnWrapper, TransformerMixin, BaseEstimator):
     """
     Base LSI module
     """
 
-    def __init__(self, corpus=None, num_topics=200, id2word=None, chunksize=20000,
+    def __init__(self, num_topics=200, id2word=None, chunksize=20000,
                  decay=1.0, onepass=True, power_iters=2, extra_samples=100):
         """
         Sklearn wrapper for LSI model. Class derived from gensim.model.LsiModel.
         """
-        self.corpus = corpus
+        self.gensim_model = None
         self.num_topics = num_topics
         self.id2word = id2word
         self.chunksize = chunksize
@@ -36,52 +38,51 @@ def __init__(self, corpus=None, num_topics=200, id2word=None, chunksize=20000,
         self.extra_samples = extra_samples
         self.power_iters = power_iters
 
-        # if 'fit' function is not used, then 'corpus' is given in init
-        if self.corpus:
-            models.LsiModel.__init__(self, corpus=self.corpus, num_topics=self.num_topics, id2word=self.id2word, chunksize=self.chunksize,
-                 decay=self.decay, onepass=self.onepass, power_iters=self.power_iters, extra_samples=self.extra_samples)
-
     def get_params(self, deep=True):
         """
         Returns all parameters as dictionary.
         """
-        return {"corpus": self.corpus, "num_topics": self.num_topics, "id2word": self.id2word,
+        return {"num_topics": self.num_topics, "id2word": self.id2word,
                 "chunksize": self.chunksize, "decay": self.decay, "onepass": self.onepass,
                 "extra_samples": self.extra_samples, "power_iters": self.power_iters}
 
     def set_params(self, **parameters):
         """
         Set all parameters.
         """
-        super(SklearnWrapperLsiModel, self).set_params(**parameters)
+        super(SklLsiModel, self).set_params(**parameters)
 
     def fit(self, X, y=None):
         """
-        For fitting corpus into the class object.
-        Calls gensim.model.LsiModel:
-        >>>gensim.models.LsiModel(corpus=corpus, num_topics=num_topics, id2word=id2word, chunksize=chunksize, decay=decay, onepass=onepass, power_iters=power_iters, extra_samples=extra_samples)
+        Fit the model according to the given training data.
+        Calls gensim.models.LsiModel
         """
         if sparse.issparse(X):
-            self.corpus = matutils.Sparse2Corpus(X)
+            corpus = matutils.Sparse2Corpus(X)
         else:
-            self.corpus = X
+            corpus = X
 
-        models.LsiModel.__init__(self, corpus=self.corpus, num_topics=self.num_topics, id2word=self.id2word, chunksize=self.chunksize,
-             decay=self.decay, onepass=self.onepass, power_iters=self.power_iters, extra_samples=self.extra_samples)
+        self.gensim_model = models.LsiModel(corpus=corpus, num_topics=self.num_topics, id2word=self.id2word, chunksize=self.chunksize,
+            decay=self.decay, onepass=self.onepass, power_iters=self.power_iters, extra_samples=self.extra_samples)
         return self
 
     def transform(self, docs):
         """
         Takes a list of documents as input ('docs').
         Returns a matrix of topic distribution for the given document bow, where a_ij
         indicates (topic_i, topic_probability_j).
+        The input `docs` should be in BOW format and can be a list of documents like : [ [(4, 1), (7, 1)], [(9, 1), (13, 1)], [(2, 1), (6, 1)] ]
+        or a single document like : [(4, 1), (7, 1)]
         """
+        if self.gensim_model is None:
+            raise NotFittedError("This model has not been fitted yet. Call 'fit' with appropriate arguments before using this method.")
+
         # The input as array of array
         check = lambda x: [x] if isinstance(x[0], tuple) else x
         docs = check(docs)
         X = [[] for i in range(0,len(docs))];
         for k,v in enumerate(docs):
-            doc_topics = self[v]
+            doc_topics = self.gensim_model[v]
             probs_docs = list(map(lambda x: x[1], doc_topics))
             # Everything should be equal in length
             if len(probs_docs) != self.num_topics:
@@ -96,4 +97,10 @@ def partial_fit(self, X):
         """
         if sparse.issparse(X):
             X = matutils.Sparse2Corpus(X)
-        self.add_documents(corpus=X)
+
+        if self.gensim_model is None:
+            self.gensim_model = models.LsiModel(num_topics=self.num_topics, id2word=self.id2word, chunksize=self.chunksize,
+                decay=self.decay, onepass=self.onepass, power_iters=self.power_iters, extra_samples=self.extra_samples)
+
+        self.gensim_model.add_documents(corpus=X)
+        return self