-
Notifications
You must be signed in to change notification settings - Fork 7
/
__init__.py
395 lines (345 loc) · 12.8 KB
/
__init__.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
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
# Copyright (c) 2014, Salesforce.com, Inc. All rights reserved.
# Copyright (c) 2015, Gamelan Labs, Inc.
# Copyright (c) 2016, Google, Inc.
# Copyright (c) 2019, Gamalon, Inc.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions
# are met:
#
# - Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
# - Redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution.
# - Neither the name of Salesforce.com nor the names of its contributors
# may be used to endorse or promote products derived from this
# software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
# FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
# COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
# INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
# BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
# OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
# ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR
# TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
# USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from __future__ import division
from collections import defaultdict
from math import gamma
try:
from itertools import izip as zip
except ImportError:
pass
import random
import sys
import numpy
import numpy.random
from numpy import pi
from scipy.spatial import cKDTree
from .utils import chi2sf
NoneType = type(None)
#: Data types for integral random variables.
#:
#: For Python 2.7, this tuple also includes `long`.
INTEGRAL_TYPES = (int, )
#: Data types for continuous random variables.
CONTINUOUS_TYPES = (float, numpy.float32, numpy.float64)
#: Data types for discrete random variables.
#:
#: For Python 2.7, this tuple also includes `long` and `basestring`.
DISCRETE_TYPES = (NoneType, bool, int, str, numpy.int32, numpy.int64)
if sys.version_info < (3, ):
# `str` is a subclass of `basestring`, so this is doing a little
# more work than is necessary, but it should not cause a problem.
DISCRETE_TYPES += (long, basestring) # noqa
INTEGRAL_TYPES += (long, ) # noqa
def seed_all(seed):
random.seed(seed)
numpy.random.seed(seed)
def get_dim(thing):
if hasattr(thing, '__len__'):
return len(thing)
else:
return 1
def print_histogram(probs, counts):
WIDTH = 60.0
max_count = max(counts)
print('{: >8} {: >8}'.format('Prob', 'Count'))
for prob, count in sorted(zip(probs, counts), reverse=True):
width = int(round(WIDTH * count / max_count))
print('{: >8.3f} {: >8d} {}'.format(prob, count, '-' * width))
def multinomial_goodness_of_fit(
probs,
counts,
total_count,
truncated=False,
plot=False):
"""
Pearson's chi^2 test, on possibly truncated data.
http://en.wikipedia.org/wiki/Pearson%27s_chi-squared_test
Returns:
p-value of truncated multinomial sample.
"""
assert len(probs) == len(counts)
assert truncated or total_count == sum(counts)
chi_squared = 0
dof = 0
if plot:
print_histogram(probs, counts)
for p, c in zip(probs, counts):
if abs(p - 1) < 1e-8:
return 1 if c == total_count else 0
assert p < 1, 'bad probability: %g' % p
if p > 0:
mean = total_count * p
variance = total_count * p * (1 - p)
assert variance > 1, \
'WARNING goodness of fit is inaccurate; use more samples'
chi_squared += (c - mean) ** 2 / variance
dof += 1
else:
print('WARNING zero probability in goodness-of-fit test')
if c > 0:
return float('inf')
if not truncated:
dof -= 1
survival = chi2sf(chi_squared, dof)
return survival
def unif01_goodness_of_fit(samples, plot=False):
"""
Bin uniformly distributed samples and apply Pearson's chi^2 test.
"""
samples = numpy.array(samples, dtype=float)
assert samples.min() >= 0.0
assert samples.max() <= 1.0
bin_count = int(round(len(samples) ** 0.333))
assert bin_count >= 7, 'WARNING imprecise test, use more samples'
probs = numpy.ones(bin_count, dtype=float) / bin_count
counts = numpy.zeros(bin_count, dtype=int)
for sample in samples:
counts[min(bin_count - 1, int(bin_count * sample))] += 1
return multinomial_goodness_of_fit(probs, counts, len(samples), plot=plot)
def exp_goodness_of_fit(
samples,
plot=False,
normalized=True,
return_dict=False):
"""
Transform exponentially distribued samples to unif01 distribution
and assess goodness of fit via binned Pearson's chi^2 test.
Inputs:
samples - a list of real-valued samples from a candidate distribution
"""
result = {}
if not normalized:
result['norm'] = numpy.mean(samples)
samples /= result['norm']
unif01_samples = numpy.exp(-samples)
result['gof'] = unif01_goodness_of_fit(unif01_samples, plot=plot)
return result if return_dict else result['gof']
def density_goodness_of_fit(
samples,
probs,
plot=False,
normalized=True,
return_dict=False):
"""
Transform arbitrary continuous samples to unif01 distribution
and assess goodness of fit via binned Pearson's chi^2 test.
Inputs:
samples - a list of real-valued samples from a distribution
probs - a list of probability densities evaluated at those samples
"""
assert len(samples) == len(probs)
assert len(samples) > 100, 'WARNING imprecision; use more samples'
pairs = list(zip(samples, probs))
pairs.sort()
samples = numpy.array([x for x, p in pairs])
probs = numpy.array([p for x, p in pairs])
density = len(samples) * numpy.sqrt(probs[1:] * probs[:-1])
gaps = samples[1:] - samples[:-1]
exp_samples = density * gaps
return exp_goodness_of_fit(exp_samples, plot, normalized, return_dict)
def volume_of_sphere(dim, radius):
assert isinstance(dim, INTEGRAL_TYPES)
return radius ** dim * pi ** (0.5 * dim) / gamma(0.5 * dim + 1)
def get_nearest_neighbor_distances(samples):
if not hasattr(samples[0], '__iter__'):
samples = numpy.array([samples]).T
distances, indices = cKDTree(samples).query(samples, k=2)
return distances[:, 1]
def vector_density_goodness_of_fit(
samples,
probs,
plot=False,
normalized=True,
return_dict=False):
"""
Transform arbitrary multivariate continuous samples
to unif01 distribution via nearest neighbor distribution [1,2,3]
and assess goodness of fit via binned Pearson's chi^2 test.
[1] http://projecteuclid.org/download/pdf_1/euclid.aop/1176993668
[2] http://arxiv.org/pdf/1006.3019v2.pdf
[3] http://en.wikipedia.org/wiki/Nearest_neighbour_distribution
Inputs:
samples - a list of real-vector-valued samples from a distribution
probs - a list of probability densities evaluated at those samples
"""
assert len(samples)
assert len(samples) == len(probs)
dim = get_dim(samples[0])
assert dim
assert len(samples) > 1000 * dim, 'WARNING imprecision; use more samples'
radii = get_nearest_neighbor_distances(samples)
density = len(samples) * numpy.array(probs)
volume = volume_of_sphere(dim, radii)
exp_samples = density * volume
return exp_goodness_of_fit(exp_samples, plot, normalized, return_dict)
def trivial_density_goodness_of_fit(
samples,
probs,
plot=False,
normalized=True,
return_dict=False):
assert len(samples)
assert all(sample == samples[0] for sample in samples)
result = {'gof': 1.0}
if not normalized:
result['norm'] = probs[0]
if return_dict:
return result
else:
return result['gof']
def auto_density_goodness_of_fit(
samples,
probs,
plot=False,
normalized=True,
return_dict=False):
"""
Dispatch on sample dimention and delegate to one of:
- density_goodness_of_fit
- vector_density_goodness_of_fit
- trivial_density_goodness_of_fit
"""
assert len(samples)
dim = get_dim(samples[0])
if dim == 0:
fun = trivial_density_goodness_of_fit
elif dim == 1:
fun = density_goodness_of_fit
if hasattr(samples[0], '__len__'):
samples = [sample[0] for sample in samples]
else:
fun = vector_density_goodness_of_fit
return fun(samples, probs, plot, normalized, return_dict)
def discrete_goodness_of_fit(
samples,
probs_dict,
truncate_beyond=8,
plot=False,
normalized=True):
"""
Transform arbitrary discrete data to multinomial
and assess goodness of fit via Pearson's chi^2 test.
"""
if not normalized:
norm = sum(probs_dict.values())
probs_dict = {i: p / norm for i, p in probs_dict.items()}
counts = defaultdict(lambda: 0)
for sample in samples:
assert sample in probs_dict
counts[sample] += 1
items = [(prob, counts.get(i, 0)) for i, prob in probs_dict.items()]
items.sort(reverse=True)
truncated = (truncate_beyond and truncate_beyond < len(items))
if truncated:
items = items[:truncate_beyond]
probs = [prob for prob, count in items]
counts = [count for prob, count in items]
assert sum(counts) > 100, 'WARNING imprecision; use more samples'
return multinomial_goodness_of_fit(
probs,
counts,
len(samples),
truncated=truncated,
plot=plot)
def split_discrete_continuous(data):
"""
Convert arbitrary data to a pair `(discrete, continuous)`
where `discrete` is hashable and `continuous` is a list of floats.
"""
if isinstance(data, DISCRETE_TYPES):
return data, []
elif isinstance(data, CONTINUOUS_TYPES):
return None, [data]
elif isinstance(data, (tuple, list)):
discrete = []
continuous = []
for part in data:
d, c = split_discrete_continuous(part)
discrete.append(d)
continuous += c
return tuple(discrete), continuous
elif isinstance(data, numpy.ndarray):
assert data.dtype in [numpy.float64, numpy.float32]
return (None,) * len(data), list(map(float, data))
else:
raise TypeError(
'split_discrete_continuous does not accept {} of type {}'.format(
repr(data), str(type(data))))
def mixed_density_goodness_of_fit(samples, probs, plot=False, normalized=True):
"""
Test general mixed discrete+continuous datatypes by
(1) testing the continuous part conditioned on each discrete value
(2) testing the discrete part marginalizing over the continuous part
(3) testing the estimated total probability (if normalized = True)
Inputs:
samples - a list of plain-old-data samples from a distribution
probs - a list of probability densities evaluated at those samples
"""
assert len(samples)
discrete_samples = []
strata = defaultdict(lambda: ([], []))
for sample, prob in zip(samples, probs):
d, c = split_discrete_continuous(sample)
discrete_samples.append(d)
samples, probs = strata[d]
samples.append(c)
probs.append(prob)
# Continuous part
gofs = []
discrete_probs = {}
for key, (samples, probs) in strata.items():
result = auto_density_goodness_of_fit(
samples,
probs,
plot=plot,
normalized=False,
return_dict=True)
gofs.append(result['gof'])
discrete_probs[key] = result['norm']
# Discrete part
if len(strata) > 1:
gofs.append(discrete_goodness_of_fit(
discrete_samples,
discrete_probs,
plot=plot,
normalized=False))
# Normalization
if normalized:
norm = sum(discrete_probs.values())
discrete_counts = [len(samples) for samples, _ in strata.values()]
norm_variance = sum(1.0 / count for count in discrete_counts)
dof = len(discrete_counts)
chi_squared = (1 - norm) ** 2 / norm_variance
gofs.append(chi2sf(chi_squared, dof))
if plot:
print('norm = {:.4g} +- {:.4g}'.format(norm, norm_variance ** 0.5))
print(' = {}'.format(
' + '.join(map('{:.4g}'.format, discrete_probs.values()))))
return min(gofs)