|
1
|
|
|
import numpy as np |
|
2
|
|
|
import pandas as pd |
|
3
|
|
|
import unittest |
|
4
|
|
|
from ..clean import data_cleaning, drop_missing, convert_datatypes, pool_duplicate_subsets |
|
5
|
|
|
|
|
6
|
|
|
|
|
7
|
|
|
class Test_drop_missing(unittest.TestCase): |
|
8
|
|
|
@classmethod |
|
9
|
|
|
def setUpClass(cls): |
|
10
|
|
|
cls.df_data_drop = pd.DataFrame( |
|
11
|
|
|
[ |
|
12
|
|
|
[np.nan, np.nan, np.nan, np.nan, np.nan], |
|
13
|
|
|
[pd.NA, pd.NA, pd.NA, pd.NA, pd.NA], |
|
14
|
|
|
[pd.NA, "b", "c", "d", "e"], |
|
15
|
|
|
[pd.NA, 6, 7, 8, 9], |
|
16
|
|
|
[pd.NA, 2, 3, 4, pd.NA], |
|
17
|
|
|
[pd.NA, 6, 7, pd.NA, pd.NA], |
|
18
|
|
|
], |
|
19
|
|
|
columns=["c1", "c2", "c3", "c4", "c5"], |
|
20
|
|
|
) |
|
21
|
|
|
|
|
22
|
|
|
def test_drop_missing(self): |
|
23
|
|
|
self.assertEqual(drop_missing(self.df_data_drop).shape, (4, 4)) |
|
24
|
|
|
|
|
25
|
|
|
# Drop further columns based on threshold |
|
26
|
|
|
self.assertEqual(drop_missing(self.df_data_drop, drop_threshold_cols=0.5).shape, (4, 3)) |
|
27
|
|
|
self.assertEqual( |
|
28
|
|
|
drop_missing(self.df_data_drop, drop_threshold_cols=0.5, col_exclude=["c1"]).shape, (4, 4) |
|
29
|
|
|
) |
|
30
|
|
|
self.assertEqual(drop_missing(self.df_data_drop, drop_threshold_cols=0.49).shape, (4, 2)) |
|
31
|
|
|
self.assertEqual(drop_missing(self.df_data_drop, drop_threshold_cols=0).shape, (0, 0)) |
|
32
|
|
|
|
|
33
|
|
|
# Drop further rows based on threshold |
|
34
|
|
|
self.assertEqual(drop_missing(self.df_data_drop, drop_threshold_rows=0.67).shape, (4, 4)) |
|
35
|
|
|
self.assertEqual(drop_missing(self.df_data_drop, drop_threshold_rows=0.5).shape, (4, 4)) |
|
36
|
|
|
self.assertEqual(drop_missing(self.df_data_drop, drop_threshold_rows=0.49).shape, (3, 4)) |
|
37
|
|
|
self.assertEqual(drop_missing(self.df_data_drop, drop_threshold_rows=0.25).shape, (3, 4)) |
|
38
|
|
|
self.assertEqual(drop_missing(self.df_data_drop, drop_threshold_rows=0.24).shape, (2, 4)) |
|
39
|
|
|
self.assertEqual( |
|
40
|
|
|
drop_missing(self.df_data_drop, drop_threshold_rows=0.24, col_exclude=["c1"]).shape, (2, 5) |
|
41
|
|
|
) |
|
42
|
|
|
self.assertEqual( |
|
43
|
|
|
drop_missing(self.df_data_drop, drop_threshold_rows=0.24, col_exclude=["c2"]).shape, (2, 4) |
|
44
|
|
|
) |
|
45
|
|
|
self.assertEqual( |
|
46
|
|
|
drop_missing(self.df_data_drop, drop_threshold_rows=0.51, col_exclude=["c1"]).shape, (3, 5) |
|
47
|
|
|
) |
|
48
|
|
|
|
|
49
|
|
|
|
|
50
|
|
|
class Test_data_cleaning(unittest.TestCase): |
|
51
|
|
|
@classmethod |
|
52
|
|
|
def setUpClass(cls): |
|
53
|
|
|
cls.df_data_cleaning = pd.DataFrame( |
|
54
|
|
|
[ |
|
55
|
|
|
[np.nan, np.nan, np.nan, np.nan, np.nan], |
|
56
|
|
|
[pd.NA, pd.NA, pd.NA, pd.NA, pd.NA], |
|
57
|
|
|
[pd.NA, "b", "c", "d", "e"], |
|
58
|
|
|
[pd.NA, 6, 7, 8, 9], |
|
59
|
|
|
[pd.NA, 2, 3, 4, pd.NA], |
|
60
|
|
|
[pd.NA, 6, 7, pd.NA, pd.NA], |
|
61
|
|
|
], |
|
62
|
|
|
columns=["c1", "c2", "c3", "c4", "c5"], |
|
63
|
|
|
) |
|
64
|
|
|
|
|
65
|
|
|
def test_data_cleaning(self): |
|
66
|
|
|
self.assertEqual(data_cleaning(self.df_data_cleaning).shape, (4, 4)) |
|
67
|
|
|
|
|
68
|
|
|
|
|
69
|
|
|
class Test_convert_dtypes(unittest.TestCase): |
|
70
|
|
|
@classmethod |
|
71
|
|
|
def setUpClass(cls): |
|
72
|
|
|
cls.df_data_convert = pd.DataFrame( |
|
73
|
|
|
[ |
|
74
|
|
|
[1, 7.0, "y", "x", pd.NA, "v"], |
|
75
|
|
|
[3, 8.0, "d", "e", pd.NA, "v"], |
|
76
|
|
|
[5, 7.0, "o", "z", pd.NA, "v"], |
|
77
|
|
|
[1, 7.0, "u", "f", pd.NA, "p"], |
|
78
|
|
|
[1, 7.0, "u", "f", pd.NA, "p"], |
|
79
|
|
|
[2, 7.0, "g", "a", pd.NA, "p"], |
|
80
|
|
|
] |
|
81
|
|
|
) |
|
82
|
|
|
|
|
83
|
|
|
def test_convert_dtypes(self): |
|
84
|
|
|
expected_results = ["Int8", "Float32", "string", "string", "category", "category"] |
|
85
|
|
|
for i, _ in enumerate(expected_results): |
|
86
|
|
|
self.assertEqual( |
|
87
|
|
|
convert_datatypes(self.df_data_convert, cat_threshold=0.4).dtypes[i], expected_results[i] |
|
88
|
|
|
) |
|
89
|
|
|
|
|
90
|
|
|
expected_results = ["Int8", "Float32", "string", "string", "object", "string"] |
|
91
|
|
|
for i, _ in enumerate(expected_results): |
|
92
|
|
|
self.assertEqual(convert_datatypes(self.df_data_convert).dtypes[i], expected_results[i]) |
|
93
|
|
|
|
|
94
|
|
|
expected_results = ["Int8", "Float32", "string", "string", "object", "category"] |
|
95
|
|
|
for i, _ in enumerate(expected_results): |
|
96
|
|
|
self.assertEqual( |
|
97
|
|
|
convert_datatypes(self.df_data_convert, cat_threshold=0.5, cat_exclude=[4]).dtypes[i], |
|
98
|
|
|
expected_results[i], |
|
99
|
|
|
) |
|
100
|
|
|
|
|
101
|
|
|
expected_results = ["Int8", "Float32", "string", "category", "object", "category"] |
|
102
|
|
|
for i, _ in enumerate(expected_results): |
|
103
|
|
|
self.assertEqual( |
|
104
|
|
|
convert_datatypes(self.df_data_convert, cat_threshold=0.95, cat_exclude=[2, 4]).dtypes[i], |
|
105
|
|
|
expected_results[i], |
|
106
|
|
|
) |
|
107
|
|
|
|
|
108
|
|
|
expected_results = ["Int8", "Float32", "string", "string", "object", "string"] |
|
109
|
|
|
for i, _ in enumerate(expected_results): |
|
110
|
|
|
self.assertEqual( |
|
111
|
|
|
convert_datatypes( |
|
112
|
|
|
self.df_data_convert, category=False, cat_threshold=0.95, cat_exclude=[2, 4] |
|
113
|
|
|
).dtypes[i], |
|
114
|
|
|
expected_results[i], |
|
115
|
|
|
) |
|
116
|
|
|
|
|
117
|
|
|
|
|
118
|
|
|
class Test_pool_duplicate_subsets(unittest.TestCase): |
|
119
|
|
|
@classmethod |
|
120
|
|
|
def setUpClass(cls): |
|
121
|
|
|
cls.df_data_subsets = pd.DataFrame( |
|
122
|
|
|
[ |
|
123
|
|
|
[1, 7, "d", "x", pd.NA, "v"], |
|
124
|
|
|
[1, 8, "d", "e", pd.NA, "v"], |
|
125
|
|
|
[2, 7, "g", "z", pd.NA, "v"], |
|
126
|
|
|
[1, 7, "u", "f", pd.NA, "p"], |
|
127
|
|
|
[1, 7, "u", "z", pd.NA, "p"], |
|
128
|
|
|
[2, 7, "g", "z", pd.NA, "p"], |
|
129
|
|
|
] |
|
130
|
|
|
) |
|
131
|
|
|
|
|
132
|
|
|
def test_pool_duplicate_subsets(self): |
|
133
|
|
|
self.assertEqual(pool_duplicate_subsets(self.df_data_subsets).shape, (6, 3)) |
|
134
|
|
|
self.assertEqual(pool_duplicate_subsets(self.df_data_subsets, col_dupl_thresh=1).shape, (6, 6)) |
|
135
|
|
|
self.assertEqual(pool_duplicate_subsets(self.df_data_subsets, subset_thresh=0).shape, (6, 2)) |
|
136
|
|
|
|
akanz1 /
klib
