|
1
|
|
|
import unittest |
|
2
|
|
|
|
|
3
|
|
|
import pandas as pd |
|
4
|
|
|
|
|
5
|
|
|
from klib.preprocess import train_dev_test_split |
|
6
|
|
|
|
|
7
|
|
|
|
|
8
|
|
|
class Test_train_dev_test_split(unittest.TestCase): |
|
9
|
|
|
@classmethod |
|
10
|
|
|
def setUpClass(cls): |
|
11
|
|
|
cls.data_split = pd.DataFrame( |
|
12
|
|
|
[ |
|
13
|
|
|
[1, 2, 3, 4, "a"], |
|
14
|
|
|
[2, 4, 5, 6, "b"], |
|
15
|
|
|
[3, 4, 2, 1, "c"], |
|
16
|
|
|
[4, 0, 3, 4, "a"], |
|
17
|
|
|
[5, 4, 5, 6, "b"], |
|
18
|
|
|
[6, 4, 2, 1, "c"], |
|
19
|
|
|
[7, 0, 3, 4, "a"], |
|
20
|
|
|
[8, 4, 5, 6, "b"], |
|
21
|
|
|
[9, 4, 2, 1, "c"], |
|
22
|
|
|
[10, 2, 1, 5, "b"], |
|
23
|
|
|
], |
|
24
|
|
|
columns=["Col1", "Col2", "Col3", "Col4", "Col5"], |
|
25
|
|
|
) |
|
26
|
|
|
cls.data_target = pd.Series([1, 0, 1, 0, 0, 1, 1, 0, 1, 1]) |
|
27
|
|
|
|
|
28
|
|
View Code Duplication |
def test_train_dev_test_split_col(self): |
|
29
|
|
|
# Test the propper splitting in train, dev and test sets |
|
30
|
|
|
|
|
31
|
|
|
expected_results = [(8, 4), (1, 4), (1, 4), (8,), (1,), (1,)] |
|
32
|
|
|
for i, _ in enumerate(expected_results): |
|
33
|
|
|
self.assertEqual( |
|
34
|
|
|
train_dev_test_split(self.data_split, "Col2", random_state=1234)[ |
|
35
|
|
|
i |
|
36
|
|
|
].shape, |
|
37
|
|
|
expected_results[i], |
|
38
|
|
|
) |
|
39
|
|
|
|
|
40
|
|
|
expected_results = [(8, 4), (2, 4), (8,), (2,)] |
|
41
|
|
|
for i, _ in enumerate(expected_results): |
|
42
|
|
|
self.assertEqual( |
|
43
|
|
|
train_dev_test_split( |
|
44
|
|
|
self.data_split, target="Col2", dev_size=0, test_size=0.2 |
|
45
|
|
|
)[i].shape, |
|
46
|
|
|
expected_results[i], |
|
47
|
|
|
) |
|
48
|
|
|
|
|
49
|
|
|
expected_results = [(5, 4), (5, 4), (5,), (5,)] |
|
50
|
|
|
for i, _ in enumerate(expected_results): |
|
51
|
|
|
self.assertEqual( |
|
52
|
|
|
train_dev_test_split( |
|
53
|
|
|
self.data_split, target="Col2", dev_size=0.5, test_size=0 |
|
54
|
|
|
)[i].shape, |
|
55
|
|
|
expected_results[i], |
|
56
|
|
|
) |
|
57
|
|
|
|
|
58
|
|
View Code Duplication |
def test_train_dev_test_split_series(self): |
|
59
|
|
|
# Test the propper splitting in train, dev and test sets |
|
60
|
|
|
|
|
61
|
|
|
expected_results = [(6, 5), (2, 5), (2, 5), (6,), (2,), (2,)] |
|
62
|
|
|
for i, _ in enumerate(expected_results): |
|
63
|
|
|
self.assertEqual( |
|
64
|
|
|
train_dev_test_split( |
|
65
|
|
|
self.data_split, |
|
66
|
|
|
target=self.data_target, |
|
67
|
|
|
dev_size=0.2, |
|
68
|
|
|
test_size=0.2, |
|
69
|
|
|
)[i].shape, |
|
70
|
|
|
expected_results[i], |
|
71
|
|
|
) |
|
72
|
|
|
|
|
73
|
|
|
expected_results = [(8, 5), (2, 5), (8,), (2,)] |
|
74
|
|
|
for i, _ in enumerate(expected_results): |
|
75
|
|
|
self.assertEqual( |
|
76
|
|
|
train_dev_test_split( |
|
77
|
|
|
self.data_split, target=self.data_target, dev_size=0, test_size=0.2 |
|
78
|
|
|
)[i].shape, |
|
79
|
|
|
expected_results[i], |
|
80
|
|
|
) |
|
81
|
|
|
|
|
82
|
|
|
expected_results = [(5, 5), (5, 5), (5,), (5,)] |
|
83
|
|
|
for i, _ in enumerate(expected_results): |
|
84
|
|
|
self.assertEqual( |
|
85
|
|
|
train_dev_test_split( |
|
86
|
|
|
self.data_split, target=self.data_target, dev_size=0.5, test_size=0 |
|
87
|
|
|
)[i].shape, |
|
88
|
|
|
expected_results[i], |
|
89
|
|
|
) |
|
90
|
|
|
|
akanz1 /
klib
