|
1
|
|
|
import numpy as np |
|
2
|
|
|
import pandas as pd |
|
3
|
|
|
import unittest |
|
4
|
|
|
from klib.utils import _corr_selector |
|
5
|
|
|
from klib.utils import _drop_duplicates |
|
6
|
|
|
from klib.utils import _missing_vals |
|
7
|
|
|
from klib.utils import _validate_input_0_1 |
|
8
|
|
|
from klib.utils import _validate_input_bool |
|
9
|
|
|
|
|
10
|
|
|
if __name__ == '__main__': |
|
11
|
|
|
unittest.main() |
|
12
|
|
|
|
|
13
|
|
|
|
|
14
|
|
|
class Test__corr_selector(unittest.TestCase): |
|
15
|
|
|
|
|
16
|
|
|
@classmethod |
|
17
|
|
|
def setUpClass(cls): |
|
18
|
|
|
cls.df_data_corr = pd.DataFrame([[1, 7, 2, 2, 4, 7], |
|
19
|
|
|
[3, 8, 3, 3, 7, 1], |
|
20
|
|
|
[5, 7, 9, 5, 1, 4], |
|
21
|
|
|
[1, 7, 8, 6, 1, 8], |
|
22
|
|
|
[1, 7, 5, 6, 2, 6], |
|
23
|
|
|
[2, 7, 3, 3, 5, 3]]) |
|
24
|
|
|
|
|
25
|
|
|
cls.target = pd.Series([1, 2, 4, 7, 4, 2]) |
|
26
|
|
|
|
|
27
|
|
|
def test__corr_selector_matrix(self): |
|
28
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corr()).shape, (6, 6)) |
|
29
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corr(), split='pos').isna().sum().sum(), 18) |
|
30
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corr(), split='pos', threshold=0.5).isna().sum().sum(), 26) |
|
31
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corr(), split='neg', threshold=-0.75).isna().sum().sum(), 32) |
|
32
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corr(), split='high', threshold=0.15).isna().sum().sum(), 4) |
|
33
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corr(), split='low', threshold=0.85).isna().sum().sum(), 6) |
|
34
|
|
|
|
|
35
|
|
|
def test__corr_selector_label(self): |
|
36
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corrwith(self.target)).shape, (6, )) |
|
37
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corrwith(self.target), split='pos').isna().sum(), 3) |
|
38
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corrwith( |
|
39
|
|
|
self.target), split='pos', threshold=0.8).isna().sum(), 4) |
|
40
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corrwith( |
|
41
|
|
|
self.target), split='neg', threshold=-0.7).isna().sum(), 5) |
|
42
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corrwith( |
|
43
|
|
|
self.target), split='high', threshold=0.2).isna().sum(), 1) |
|
44
|
|
|
self.assertEqual(_corr_selector(self.df_data_corr.corrwith( |
|
45
|
|
|
self.target), split='low', threshold=0.8).isna().sum(), 2) |
|
46
|
|
|
|
|
47
|
|
|
|
|
48
|
|
|
class Test__drop_duplicates(unittest.TestCase): |
|
49
|
|
|
|
|
50
|
|
|
@classmethod |
|
51
|
|
|
def setUpClass(cls): |
|
52
|
|
|
cls.data_dupl_df = pd.DataFrame([[pd.NA, pd.NA, pd.NA, pd.NA], |
|
53
|
|
|
[1, 2, 3, 4], |
|
54
|
|
|
[1, 2, 3, 4], |
|
55
|
|
|
[1, 2, 3, 4], |
|
56
|
|
|
[2, 3, 4, 5], |
|
57
|
|
|
[1, 2, 3, pd.NA], |
|
58
|
|
|
[pd.NA, pd.NA, pd.NA, pd.NA]]) |
|
59
|
|
|
|
|
60
|
|
|
def test__drop_dupl(self): |
|
61
|
|
|
# Test dropping of duplicate rows |
|
62
|
|
|
self.assertAlmostEqual(_drop_duplicates(self.data_dupl_df)[0].shape, (4, 4)) |
|
63
|
|
|
# Test if the resulting DataFrame is equal to using the pandas method |
|
64
|
|
|
self.assertTrue(_drop_duplicates(self.data_dupl_df)[0].equals(self.data_dupl_df.drop_duplicates())) |
|
65
|
|
|
# Test number of duplicates |
|
66
|
|
|
self.assertEqual(len(_drop_duplicates(self.data_dupl_df)[1]), 3) |
|
67
|
|
|
|
|
68
|
|
|
|
|
69
|
|
|
class Test__missing_vals(unittest.TestCase): |
|
70
|
|
|
|
|
71
|
|
|
@classmethod |
|
72
|
|
|
def setUpClass(cls): |
|
73
|
|
|
cls.data_mv_df = pd.DataFrame([[1, np.nan, 3, 4], |
|
74
|
|
|
[None, 4, 5, None], |
|
75
|
|
|
['a', 'b', pd.NA, 'd'], |
|
76
|
|
|
[True, False, 7, pd.NaT]]) |
|
77
|
|
|
|
|
78
|
|
|
cls.data_mv_array = np.array([[1, np.nan, 3, 4], |
|
79
|
|
|
[None, 4, 5, None], |
|
80
|
|
|
['a', 'b', pd.NA, 'd'], |
|
81
|
|
|
[True, False, 7, pd.NaT]]) |
|
82
|
|
|
|
|
83
|
|
|
cls.data_mv_list = [[1, np.nan, 3, 4], |
|
84
|
|
|
[None, 4, 5, None], |
|
85
|
|
|
['a', 'b', pd.NA, 'd'], |
|
86
|
|
|
[True, False, 7, pd.NaT]] |
|
87
|
|
|
|
|
88
|
|
|
def test_mv_total(self): |
|
89
|
|
|
# Test total missing values |
|
90
|
|
|
self.assertAlmostEqual(_missing_vals(self.data_mv_df)['mv_total'], 5) |
|
91
|
|
|
self.assertAlmostEqual(_missing_vals(self.data_mv_array)['mv_total'], 5) |
|
92
|
|
|
self.assertAlmostEqual(_missing_vals(self.data_mv_list)['mv_total'], 5) |
|
93
|
|
|
|
|
94
|
|
|
def test_mv_rows(self): |
|
95
|
|
|
# Test missing values for each row |
|
96
|
|
|
expected_results = [1, 2, 1, 1] |
|
97
|
|
|
for i, _ in enumerate(expected_results): |
|
98
|
|
|
self.assertAlmostEqual(_missing_vals(self.data_mv_df)['mv_rows'][i], expected_results[i]) |
|
99
|
|
|
|
|
100
|
|
|
def test_mv_cols(self): |
|
101
|
|
|
# Test missing values for each column |
|
102
|
|
|
expected_results = [1, 1, 1, 2] |
|
103
|
|
|
for i, _ in enumerate(expected_results): |
|
104
|
|
|
self.assertAlmostEqual(_missing_vals(self.data_mv_df)['mv_cols'][i], expected_results[i]) |
|
105
|
|
|
|
|
106
|
|
|
def test_mv_rows_ratio(self): |
|
107
|
|
|
# Test missing values ratio for each row |
|
108
|
|
|
expected_results = [0.25, 0.5, 0.25, 0.25] |
|
109
|
|
|
for i, _ in enumerate(expected_results): |
|
110
|
|
|
self.assertAlmostEqual(_missing_vals(self.data_mv_df)['mv_rows_ratio'][i], expected_results[i]) |
|
111
|
|
|
|
|
112
|
|
|
# Test if missing value ratio is between 0 and 1 |
|
113
|
|
|
for i in range(len(self.data_mv_df)): |
|
114
|
|
|
self.assertTrue(0 <= _missing_vals(self.data_mv_df)['mv_rows_ratio'][i] <= 1) |
|
115
|
|
|
|
|
116
|
|
|
def test_mv_cols_ratio(self): |
|
117
|
|
|
# Test missing values ratio for each column |
|
118
|
|
|
expected_results = [1/4, 0.25, 0.25, 0.5] |
|
119
|
|
|
for i, _ in enumerate(expected_results): |
|
120
|
|
|
self.assertAlmostEqual(_missing_vals(self.data_mv_df)['mv_cols_ratio'][i], expected_results[i]) |
|
121
|
|
|
|
|
122
|
|
|
# Test if missing value ratio is between 0 and 1 |
|
123
|
|
|
for i in range(len(self.data_mv_df)): |
|
124
|
|
|
self.assertTrue(0 <= _missing_vals(self.data_mv_df)['mv_cols_ratio'][i] <= 1) |
|
125
|
|
|
|
|
126
|
|
|
|
|
127
|
|
|
class Test__validate_input(unittest.TestCase): |
|
128
|
|
|
|
|
129
|
|
|
def test__validate_input_0_1(self): |
|
130
|
|
|
with self.assertRaises(ValueError): |
|
131
|
|
|
_validate_input_0_1(-0.1, '-0.1') |
|
132
|
|
|
|
|
133
|
|
|
with self.assertRaises(ValueError): |
|
134
|
|
|
_validate_input_0_1(1.1, '1.1') |
|
135
|
|
|
|
|
136
|
|
|
def test__validate_input_bool(self): |
|
137
|
|
|
# Raises an exception if the input is not boolean |
|
138
|
|
|
with self.assertRaises(ValueError): |
|
139
|
|
|
_validate_input_bool('True', None) |
|
140
|
|
|
with self.assertRaises(ValueError): |
|
141
|
|
|
_validate_input_bool(None, None) |
|
142
|
|
|
with self.assertRaises(ValueError): |
|
143
|
|
|
_validate_input_bool(1, None) |
|
144
|
|
|
|
akanz1 /
klib
