def test_get_term_df(self):
categories, documents = get_docs_categories()
df = pd.DataFrame({'category': categories,
'text': documents})
tdm_factory = TermDocMatrixFromPandas(df,
'category',
'text',
nlp=whitespace_nlp)
term_doc_matrix = tdm_factory.build()
term_df = term_doc_matrix.get_term_freq_df()
self.assertEqual(dict(term_df.ix['speak up']),
{'??? freq': 2, 'hamlet freq': 0, 'jay-z/r. kelly freq': 1})
self.assertEqual(dict(term_df.ix['that']),
{'??? freq': 0, 'hamlet freq': 2, 'jay-z/r. kelly freq': 0})
def test_main(self):
categories, documents = get_docs_categories()
df = pd.DataFrame({'category': categories,
'text': documents})
tdm_factory = TermDocMatrixFromPandas(df,
'category',
'text',
nlp=whitespace_nlp)
term_doc_matrix = tdm_factory.build()
self.assertIsInstance(term_doc_matrix, TermDocMatrix)
self.assertEqual(set(term_doc_matrix.get_categories()),
set(['hamlet', 'jay-z/r. kelly']))
self.assertEqual(term_doc_matrix.get_num_docs(), 9)
term_doc_df = term_doc_matrix.get_term_freq_df()
self.assertEqual(term_doc_df.ix['of'].sum(), 3)
def test_main(self):
categories, documents = get_docs_categories()
df = pd.DataFrame({'category': categories,
'text': documents})
tdm_factory = TermDocMatrixFromPandas(df,
'category',
'text',
nlp=whitespace_nlp)
term_doc_matrix = tdm_factory.build()
self.assertIsInstance(term_doc_matrix, TermDocMatrix)
self.assertEqual(set(term_doc_matrix.get_categories()),
set(['hamlet','jay-z/r. kelly']))
self.assertEqual(term_doc_matrix.get_num_docs(), 9)
term_doc_df = term_doc_matrix.get_term_freq_df()
self.assertEqual(term_doc_df.ix['of'].sum(), 3)
def test_get_term_df(self):
categories, documents = get_docs_categories()
df = pd.DataFrame({'category': categories,
'text': documents})
tdm_factory = TermDocMatrixFromPandas(df,
'category',
'text',
nlp=whitespace_nlp)
term_doc_matrix = tdm_factory.build()
term_df = term_doc_matrix.get_term_freq_df()
self.assertEqual(dict(term_df.ix['speak up']),
{'??? freq': 2, 'hamlet freq': 0, 'jay-z/r. kelly freq': 1})
self.assertEqual(dict(term_df.ix['that']),
{'??? freq': 0, 'hamlet freq': 2, 'jay-z/r. kelly freq': 0})