def __init__(self,
X,
mX,
y,
term_idx_store,
category_idx_store,
metadata_idx_store,
unigram_frequency_path=None):
'''
Parameters
----------
X : csr_matrix
term document matrix
mX : csr_matrix
metadata-document matrix
y : np.array
category index array
term_idx_store : IndexStore
Term indices
category_idx_store : IndexStore
Catgory indices
metadata_idx : IndexStore
Document metadata indices
unigram_frequency_path : str or None
Path to term frequency file.
'''
TermDocMatrixWithoutCategories.__init__(self, X=X, mX=mX, term_idx_store=term_idx_store,
metadata_idx_store=metadata_idx_store,
unigram_frequency_path=unigram_frequency_path)
self._y = y
self._category_idx_store = category_idx_store
def _apply_pipeline_and_get_build_instance(self,
X_factory,
mX_factory,
df,
parse_pipeline,
term_idx_store,
metadata_idx_store):
df.apply(parse_pipeline.parse, axis=1)
X, mX = build_sparse_matrices_with_num_docs(len(df), X_factory, mX_factory)
tdm = TermDocMatrixWithoutCategories(X, mX, term_idx_store, metadata_idx_store)
return tdm
def __init__(self,
X,
mX,
y,
term_idx_store,
category_idx_store,
metadata_idx_store,
unigram_frequency_path=None):
'''
Parameters
----------
X : csr_matrix
term document matrix
mX : csr_matrix
metadata-document matrix
y : np.array
category index array
term_idx_store : IndexStore
Term indices
category_idx_store : IndexStore
Catgory indices
metadata_idx : IndexStore
Document metadata indices
unigram_frequency_path : str or None
Path to term frequency file.
'''
if all(y == y[0]):
raise CannotCreateATermDocMatrixWithASignleCategoryException(
'Documents must be labeled with more than one category. All documents were labeled '
'with category: "' + str(category_idx_store.getval(y[0])) +
'"')
TermDocMatrixWithoutCategories.__init__(
self,
X=X,
mX=mX,
term_idx_store=term_idx_store,
metadata_idx_store=metadata_idx_store,
unigram_frequency_path=unigram_frequency_path)
self._y = y
self._category_idx_store = category_idx_store