def test_fillNAs(self):
df = pd.DataFrame({'A': [11, 22, np.nan, 44]})
cfe = CategoricalFeatureExtraction()
cfe.fillNAs(df, ['A'])
self.assertEqual(df['A'][0], 11, 'Index 0 should be unchanged!')
self.assertEqual(df['A'][1], 22, 'Index 1 should be unchanged!')
self.assertTrue(df['A'][2] <= -99999999, 'Index 2 should be a large negative number!')
self.assertEqual(df['A'][3], 44, 'Index 3 should be unchanged!')
def test_convertColumns_thresholdShouldDetermineIfColumnIsOneHotEncoded(self):
CategoricalFeatureExtraction.convertColumns([self.df1, self.df2], ['animal', 'type'], one_hot_threshold=3)
#animal should be ordinal, #type should be one-hot
for df in [self.df1, self.df2]:
for onehot_col in self.type_onehot_columns:
self.assertTrue(onehot_col in df.columns)
self.assertTrue('type' not in df.columns, 'type column should have been deleted')
self.assertTrue('animal' in df.columns)
def test_convertColumnsToOneHot(self):
cfe = CategoricalFeatureExtraction()
cfe.convertColumnsToOneHot([self.df1, self.df2], ['animal'])
for df in [self.df1, self.df2]:
for onehot_col in self.animal_onehot_columns:
self.assertTrue(onehot_col in df.columns, 'one-hot column \'{0}\' is missing!'.format(onehot_col))
self.assertTrue('animal' not in df.columns, 'animal column should have been deleted!')
self.assertTrue('age' in df.columns, 'age column should not have been deleted!')
def test_convertColumnsToOrdinal(self):
cfe = CategoricalFeatureExtraction()
cfe.convertColumnsToOrdinal([self.df1, self.df2], ['age'])
for df in [self.df1, self.df2]:
print(df.columns.values)
self.assertTrue('animal' in df.columns, 'animal column should not have been deleted!')
self.assertTrue('age' in df.columns, 'age column should not have been deleted!')
self.assertEqual(self.df1['age'][0], 0, '11 should map to 0!')
self.assertEqual(self.df1['age'][1], 1, '22 should map to 1!')
self.assertEqual(self.df1['age'][2], 0, '11 should map to 0!')
self.assertEqual(self.df1['age'][3], 0, '11 should map to 0!')
self.assertEqual(self.df2['age'][0], 0, '11 should map to 0!')
self.assertEqual(self.df2['age'][1], 2, '33 should map to 2!')
self.assertEqual(self.df2['age'][2], 2, '33 should map to 2!')
def test_removeUnsharedColumns(self):
cfe = CategoricalFeatureExtraction()
df1 = pd.DataFrame({'A': [1,2,3], 'B': [2,3,4], 'C': [3,4,5]})
df2 = pd.DataFrame({'A': [1,2,3], 'B': [2,3,4]})
df3 = pd.DataFrame({'A': [1,2,3], 'C': [2,3,4]})
df4 = pd.DataFrame({'A': [1,2,3], 'D': [2,3,4]})
cfe.removeUnsharedColumns([df1, df2, df3, df4])
self.assertEqual(len(df1.columns), 1, '{0} columns found for df1 instead of 1!'.format(len(df1.columns)))
self.assertEqual(len(df2.columns), 1, '{0} columns found for df2! instead of 1'.format(len(df2.columns)))
self.assertEqual(len(df3.columns), 1, '{0} columns found for df3! instead of 1'.format(len(df3.columns)))
self.assertEqual(len(df4.columns), 1, '{0} columns found for df4! instead of 1'.format(len(df4.columns)))
self.assertEqual(df1.columns[0], 'A', 'column for df1 didn\'t contain \'A\'!')
self.assertEqual(df2.columns[0], 'A', 'column for df2 didn\'t contain \'A\'!')
self.assertEqual(df3.columns[0], 'A', 'column for df3 didn\'t contain \'A\'!')
self.assertEqual(df4.columns[0], 'A', 'column for df4 didn\'t contain \'A\'!')
