def test_to_pandas_density(self): """Should create a pandas dataframe of a denisty plot of the histogram""" hist = Histogram(bins=2) test_df = self.create_test_df() column_to_ad = test_df.select(F.col('value')) column_to_ad_2 = test_df.select(F.col('value2')) hist.add_column(column_to_ad) hist.add_column(column_to_ad_2) expected_df = pd.DataFrame({ 'value': [1.0, 0.5], 'value2': [0.5, 1.0] }).set_index([[1.75, 3.25]]) self.assertTrue(expected_df.equals(hist.to_pandas('density')))
def test_to_pandas_default(self): """Should create a pandas dataframe from the Histogram object""" hist = Histogram(bins=2) test_df = self.create_test_df() column_to_ad = test_df.select(F.col('value')) column_to_ad_2 = test_df.select(F.col('value2')) hist.add_column(column_to_ad) hist.add_column(column_to_ad_2) expected_df = pd.DataFrame({ 'value': [2, 1], 'value2': [1, 2] }).set_index([['1.00 - 2.50', '2.50 - 4.00']]) self.assertTrue(expected_df.equals(hist.to_pandas()))