forked from IntelPython/scikit-learn_bench
-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathmake_datasets.py
190 lines (160 loc) · 7.64 KB
/
make_datasets.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
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
# Copyright (C) 2018-2020 Intel Corporation
#
# SPDX-License-Identifier: MIT
import argparse
import sys
import numpy as np
from sklearn.datasets import make_classification, make_regression, make_blobs
from sklearn.utils import check_random_state
def gen_blobs(args):
X, y = make_blobs(n_samples=args.samples + args.test_samples,
n_features=args.features,
centers=None,
center_box=(-32, 32),
shuffle=True,
random_state=args.seed)
np.save(args.filex, X[:args.samples])
if args.test_samples != 0:
np.save(args.filextest, X[args.samples:])
return 0
def gen_regression(args):
rs = check_random_state(args.seed)
X, y = make_regression(n_targets=1,
n_samples=args.samples + args.test_samples,
n_features=args.features,
n_informative=args.features,
bias=rs.normal(0, 3),
random_state=rs)
np.save(args.filex, X[:args.samples])
np.save(args.filey, y[:args.samples])
if args.test_samples != 0:
np.save(args.filextest, X[args.samples:])
np.save(args.fileytest, y[args.samples:])
return 0
def gen_classification(args):
X, y = make_classification(n_samples=args.samples + args.test_samples,
n_features=args.features,
n_informative=args.features,
n_repeated=0,
n_redundant=0,
n_classes=args.classes,
random_state=args.seed)
np.save(args.filex, X[:args.samples])
np.save(args.filey, y[:args.samples])
if args.test_samples != 0:
np.save(args.filextest, X[args.samples:])
np.save(args.fileytest, y[args.samples:])
return 0
def _ch_size(n):
return n * (n + 1) // 2
def _get_cluster_centers(clusters, features):
import numpy.random_intel as nri
rs = nri.RandomState(1234, brng='SFMT19937')
cluster_centers = rs.randn(clusters, features)
cluster_centers *= np.double(clusters)
return cluster_centers
def gen_kmeans(args):
try:
import numpy.random_intel as nri
except ImportError:
raise ImportError('numpy.random_intel not found. '
'Please use Intel Distribution for Python.')
rs = nri.RandomState(args.seed, brng=('MT2203', args.node_id))
# generate centers
cluster_centers = _get_cluster_centers(args.clusters, args.features)
pvec = np.full((args.clusters,), 1.0 / args.clusters, dtype=np.double)
cluster_sizes = rs.multinomial(args.samples, pvec)
cluster_sizes_cum = cluster_sizes.cumsum()
# generate clusters around those centers
sz = 0.5
ch = rs.uniform(low=-sz, high=sz, size=(_ch_size(args.features),))
data = rs.multinormal_cholesky(cluster_centers[0], ch,
size=(args.samples + args.test_samples,))
diff_i0 = np.empty_like(cluster_centers[0])
for i in range(1, args.clusters):
np.subtract(cluster_centers[i], cluster_centers[0], out=diff_i0)
data[cluster_sizes_cum[i-1]:cluster_sizes_cum[i]] += diff_i0
j = nri.choice(range(0, args.samples), size=args.clusters, replace=False)
X_init = data[j]
X = data
times = []
import timeit
for n in range(10):
t1 = timeit.default_timer()
variances = np.var(X, axis=0)
absTol = np.mean(variances) * 1e-16
t2 = timeit.default_timer()
times.append(t2-t1)
print(f'Computing absolute threshold on this machine '
f'takes {min(times)} seconds')
np.save(args.filex, X[:args.samples])
if args.test_samples != 0:
np.save(args.filextest, X[args.samples:])
np.save(args.filei, X_init)
np.save(args.filet, absTol)
return 0
def main():
parser = argparse.ArgumentParser(
description='Dataset generator using scikit-learn')
parser.add_argument('-f', '--features', type=int, default=1000,
help='Number of features in dataset')
parser.add_argument('-s', '--samples', type=int, default=10000,
help='Number of samples in dataset')
parser.add_argument('--ts', '--test-samples', type=int, default=0,
dest='test_samples',
help='Number of test samples in dataset')
parser.add_argument('-d', '--seed', type=int, default=0,
help='Seed for random state')
subparsers = parser.add_subparsers(dest='problem')
subparsers.required = True
regr_parser = subparsers.add_parser('regression',
help='Regression data')
regr_parser.set_defaults(func=gen_regression)
regr_parser.add_argument('-x', '--filex', '--fileX', type=str,
required=True, help='Path to save matrix X')
regr_parser.add_argument('-y', '--filey', '--fileY', type=str,
required=True, help='Path to save vector y')
regr_parser.add_argument('--xt', '--filextest', '--fileXtest', type=str,
dest='filextest',
help='Path to save test matrix X')
regr_parser.add_argument('--yt', '--fileytest', '--fileYtest', type=str,
dest='fileytest',
help='Path to save test vector y')
clsf_parser = subparsers.add_parser('classification',
help='Classification data')
clsf_parser.set_defaults(func=gen_classification)
clsf_parser.add_argument('-c', '--classes', type=int, default=5,
help='Number of classes')
clsf_parser.add_argument('-x', '--filex', '--fileX', type=str,
required=True, help='Path to save matrix X')
clsf_parser.add_argument('-y', '--filey', '--fileY', type=str,
required=True,
help='Path to save label vector y')
clsf_parser.add_argument('--xt', '--filextest', '--fileXtest', type=str,
dest='filextest',
help='Path to save test matrix X')
clsf_parser.add_argument('--yt', '--fileytest', '--fileYtest', type=str,
dest='fileytest',
help='Path to save test vector y')
kmeans_parser = subparsers.add_parser('kmeans',
help='KMeans clustering data')
kmeans_parser.set_defaults(func=gen_kmeans)
kmeans_parser.add_argument('-c', '--clusters', type=int, default=10,
help='Number of clusters to generate')
kmeans_parser.add_argument('-n', '--node-id', type=int, default=0,
help='ID of member of MKL BRNG')
kmeans_parser.add_argument('-x', '--filex', '--fileX', type=str,
required=True, help='Path to save matrix X')
kmeans_parser.add_argument('--xt', '--filextest', '--fileXtest', type=str,
dest='filextest',
help='Path to test save matrix X')
kmeans_parser.add_argument('-i', '--filei', '--fileI', type=str,
required=True,
help='Path to save initial cluster centers')
kmeans_parser.add_argument('-t', '--filet', '--fileT', type=str,
required=True,
help='Path to save absolute threshold')
args = parser.parse_args()
return args.func(args)
if __name__ == '__main__':
sys.exit(main())