File: bench_sample_without_replacement.py

package info (click to toggle)
scikit-learn 0.18-5
  • links: PTS, VCS
  • area: main
  • in suites: stretch
  • size: 71,040 kB
  • ctags: 91,142
  • sloc: python: 97,257; ansic: 8,360; cpp: 5,649; makefile: 242; sh: 238
file content (207 lines) | stat: -rw-r--r-- 8,008 bytes parent folder | download | duplicates (2)
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
"""
Benchmarks for sampling without replacement of integer.

"""
from __future__ import division
from __future__ import print_function

import gc
import sys
import optparse
from datetime import datetime
import operator

import matplotlib.pyplot as plt
import numpy as np
import random

from sklearn.externals.six.moves import xrange
from sklearn.utils.random import sample_without_replacement


def compute_time(t_start, delta):
    mu_second = 0.0 + 10 ** 6  # number of microseconds in a second

    return delta.seconds + delta.microseconds / mu_second


def bench_sample(sampling, n_population, n_samples):
    gc.collect()
    # start time
    t_start = datetime.now()
    sampling(n_population, n_samples)
    delta = (datetime.now() - t_start)
    # stop time
    time = compute_time(t_start, delta)
    return time

if __name__ == "__main__":
    ###########################################################################
    # Option parser
    ###########################################################################
    op = optparse.OptionParser()
    op.add_option("--n-times",
                  dest="n_times", default=5, type=int,
                  help="Benchmark results are average over n_times experiments")

    op.add_option("--n-population",
                  dest="n_population", default=100000, type=int,
                  help="Size of the population to sample from.")

    op.add_option("--n-step",
                  dest="n_steps", default=5, type=int,
                  help="Number of step interval between 0 and n_population.")

    default_algorithms = "custom-tracking-selection,custom-auto," \
                         "custom-reservoir-sampling,custom-pool,"\
                         "python-core-sample,numpy-permutation"

    op.add_option("--algorithm",
                  dest="selected_algorithm",
                  default=default_algorithms,
                  type=str,
                  help="Comma-separated list of transformer to benchmark. "
                       "Default: %default. \nAvailable: %default")

    # op.add_option("--random-seed",
    #               dest="random_seed", default=13, type=int,
    #               help="Seed used by the random number generators.")

    (opts, args) = op.parse_args()
    if len(args) > 0:
        op.error("this script takes no arguments.")
        sys.exit(1)

    selected_algorithm = opts.selected_algorithm.split(',')
    for key in selected_algorithm:
        if key not in default_algorithms.split(','):
            raise ValueError("Unknown sampling algorithm \"%s\" not in (%s)."
                             % (key, default_algorithms))

    ###########################################################################
    # List sampling algorithm
    ###########################################################################
    # We assume that sampling algorithm has the following signature:
    #   sample(n_population, n_sample)
    #
    sampling_algorithm = {}

    ###########################################################################
    # Set Python core input
    sampling_algorithm["python-core-sample"] = \
        lambda n_population, n_sample: \
            random.sample(xrange(n_population), n_sample)

   ###########################################################################
    # Set custom automatic method selection
    sampling_algorithm["custom-auto"] = \
        lambda n_population, n_samples, random_state=None: \
            sample_without_replacement(n_population,
                                       n_samples,
                                       method="auto",
                                       random_state=random_state)

    ###########################################################################
    # Set custom tracking based method
    sampling_algorithm["custom-tracking-selection"] = \
        lambda n_population, n_samples, random_state=None: \
            sample_without_replacement(n_population,
                                       n_samples,
                                       method="tracking_selection",
                                       random_state=random_state)

    ###########################################################################
    # Set custom reservoir based method
    sampling_algorithm["custom-reservoir-sampling"] = \
        lambda n_population, n_samples, random_state=None: \
            sample_without_replacement(n_population,
                                       n_samples,
                                       method="reservoir_sampling",
                                       random_state=random_state)

    ###########################################################################
    # Set custom reservoir based method
    sampling_algorithm["custom-pool"] = \
        lambda n_population, n_samples, random_state=None: \
            sample_without_replacement(n_population,
                                       n_samples,
                                       method="pool",
                                       random_state=random_state)

    ###########################################################################
    # Numpy permutation based
    sampling_algorithm["numpy-permutation"] = \
        lambda n_population, n_sample: \
            np.random.permutation(n_population)[:n_sample]

    ###########################################################################
    # Remove unspecified algorithm
    sampling_algorithm = dict((key, value)
                              for key, value in sampling_algorithm.items()
                              if key in selected_algorithm)

    ###########################################################################
    # Perform benchmark
    ###########################################################################
    time = {}
    n_samples = np.linspace(start=0, stop=opts.n_population,
        num=opts.n_steps).astype(np.int)

    ratio = n_samples / opts.n_population

    print('Benchmarks')
    print("===========================")

    for name in sorted(sampling_algorithm):
        print("Perform benchmarks for %s..." % name, end="")
        time[name] = np.zeros(shape=(opts.n_steps, opts.n_times))

        for step in xrange(opts.n_steps):
            for it in xrange(opts.n_times):
                time[name][step, it] = bench_sample(sampling_algorithm[name],
                                                      opts.n_population,
                                                      n_samples[step])

        print("done")

    print("Averaging results...", end="")
    for name in sampling_algorithm:
        time[name] = np.mean(time[name], axis=1)
    print("done\n")

    # Print results
    ###########################################################################
    print("Script arguments")
    print("===========================")
    arguments = vars(opts)
    print("%s \t | %s " % ("Arguments".ljust(16),
                           "Value".center(12),))
    print(25 * "-" + ("|" + "-" * 14) * 1)
    for key, value in arguments.items():
        print("%s \t | %s " % (str(key).ljust(16),
                               str(value).strip().center(12)))
    print("")

    print("Sampling algorithm performance:")
    print("===============================")
    print("Results are averaged over %s repetition(s)." % opts.n_times)
    print("")

    fig = plt.figure('scikit-learn sample w/o replacement benchmark results')
    plt.title("n_population = %s, n_times = %s" %
              (opts.n_population, opts.n_times))
    ax = fig.add_subplot(111)
    for name in sampling_algorithm:
        ax.plot(ratio, time[name], label=name)

    ax.set_xlabel('ratio of n_sample / n_population')
    ax.set_ylabel('Time (s)')
    ax.legend()

    # Sort legend labels
    handles, labels = ax.get_legend_handles_labels()
    hl = sorted(zip(handles, labels), key=operator.itemgetter(1))
    handles2, labels2 = zip(*hl)
    ax.legend(handles2, labels2, loc=0)

    plt.show()