-
Notifications
You must be signed in to change notification settings - Fork 1
/
extract_dataset_info.py
73 lines (48 loc) · 2.06 KB
/
extract_dataset_info.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
import datasets
import numpy as np
import pandas as pd
from collections import Counter
from pathlib import Path
from sklearn.neighbors import NearestNeighbors
from tqdm import tqdm
def extract(k=5, verbose=True):
rows = []
columns = ['Name', 'DI', 'IR', 'Samples', 'Features']
for name in tqdm(datasets.names()):
dataset = datasets.load(name)
(X_train, y_train), (X_test, y_test) = dataset[0][0], dataset[0][1]
X = np.concatenate([X_train, X_test])
y = np.concatenate([y_train, y_test])
n_samples = X.shape[0]
n_features = X.shape[1]
majority_class = Counter(y).most_common()[0][0]
n_majority_samples = Counter(y).most_common()[0][1]
n_minority_samples = Counter(y).most_common()[1][1]
imbalance_ratio = np.round(n_majority_samples / n_minority_samples, 2)
knn = NearestNeighbors(k + 1).fit(X)
difficulty_coefficients = []
for X_i, y_i in zip(X, y):
if y_i == majority_class:
continue
else:
indices = knn.kneighbors([X_i], return_distance=False)[0, 1:]
n_majority_neighbors = sum(y[indices] == majority_class)
difficulty_coefficients.append(n_majority_neighbors / k)
difficulty_index = np.round(np.mean(difficulty_coefficients), 3)
rows.append([name, difficulty_index, imbalance_ratio, n_samples, n_features])
df = pd.DataFrame(rows, columns=columns)
df = df.sort_values('DI')
df.to_csv(Path(__file__).parent / 'results' / 'dataset_info.csv', index=False)
if verbose:
for column in ['DI', 'IR']:
df[column] = df[column].map(lambda x: f'{x:.2f}')
for i in range(30):
row = [str(df.iloc[i][c]) for c in columns]
if i + 30 < len(df):
row += [str(df.iloc[i + 30][c]) for c in columns]
else:
row += ['' for _ in columns]
print(' & '.join(row).replace('_', '\\_') + ' \\\\')
return df
if __name__ == '__main__':
extract(verbose=True)