def test_df_extra_ev_value_error():
"""to_matrix should raise NotImplementError if given a DataFrame and extra
evidence (for now)."""
lr = LogisticRegression()
source_list = ['reach', 'sparser', 'signor']
cs = CountsScorer(lr, source_list)
cs.to_matrix(test_df, extra_evidence=[[5]])
def test_matrix_to_matrix():
"""Check that we get a matrix back when passed to to_matrix."""
lr = LogisticRegression()
source_list = ['reach', 'sparser', 'signor']
cw = CountsScorer(lr, source_list)
# Train on statement data
stmt_arr = cw.to_matrix(test_df)
assert cw.to_matrix(stmt_arr) is stmt_arr, \
'If passed a numpy array to_matrix should return it back.'
def test_use_members_with_stmts():
"""Check that we can set use_num_members when passing statements."""
lr = LogisticRegression()
source_list = ['reach', 'sparser', 'signor']
cw = CountsScorer(lr, source_list, use_num_members=True)
x_arr = cw.to_matrix(test_stmts)
assert x_arr.shape == (len(test_stmts), len(source_list)+1), \
'stmt matrix dimensions should match test stmts plus num_members'
def test_use_members_with_df():
"""Check that we can't set use_num_members when passing a DataFrame."""
lr = LogisticRegression()
source_list = ['reach', 'sparser', 'signor']
cw = CountsScorer(lr, source_list, use_num_members=True)
# This should error because stmt DataFrame doesn't contain num_members
# info
stmt_arr = cw.to_matrix(test_df)