def test_pearson(self):
data = {
'C01': [1, 2, 3, 4],
'C02': [1, 2, 3, 4],
'C03': [4, 3, 2, 1],
'C04': [4, 3, 2, 1],
'C05': [1, 2, 3, 4],
'C06': [4, 3, 2, 1]
}
df = pandas.DataFrame(
data, columns=['C01', 'C02', 'C03', 'C04', 'C05', 'C06'])
corr, sign, columns, nominal_columns, inf_nan, single_value_columns = compute_correlations(
df)
print(corr)
print(sign)
# assert diagonal = 1.0
self.assertEqual(corr.at['C01', 'C01'], 1.0)
self.assertEqual(corr.at['C02', 'C02'], 1.0)
self.assertEqual(corr.at['C03', 'C03'], 1.0)
self.assertEqual(corr.at['C04', 'C04'], 1.0)
# assert some others also = 1.0
self.assertEqual(corr.at['C01', 'C02'], 1.0)
self.assertEqual(corr.at['C02', 'C01'], 1.0)
self.assertEqual(corr.at['C02', 'C03'], 1.0)
self.assertEqual(corr.at['C03', 'C02'], 1.0)
self.assertEqual(corr.at['C03', 'C04'], 1.0)
self.assertEqual(corr.at['C03', 'C04'], 1.0)
"SpatMax": "SMax",
"SpatAvg": "SAvg",
"Coverage": "Cov",
"TempDist": "TDist",
"SpatDist": "SDist",
'Strasse': "Str"
})
baysis_encoded = baysis_encoded.drop(
columns=["TempGL", "SpatGL", "TempIL", "SpatIL"])
# Calculate with Cramers 's V
results = None # To make sure that no old data is reused
results = compute_correlations(baysis_encoded,
columns_nominal=nominal_columns,
columns_dichotomous=dichotomous_columns,
columns_ordinal=ordinal_columns,
bias_correction=False)
# Plot correlation matrix
plot_correlation(results.get('correlation'),
results.get('columns'),
nominal_columns,
dichotomous_columns,
ordinal_columns,
results.get('inf_nan_corr'),
results.get('columns_single_value'),
save=save_plot,
filepath=plot_path + file_prefix + '_corr_cramers.pdf',
show=show_plot,
figsize=(18, 15))
