-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmultivariate_power_analysis_comparison.py
164 lines (132 loc) · 7.35 KB
/
multivariate_power_analysis_comparison.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
import matplotlib.pyplot as plt
import numpy as np
import os
import seaborn as sns
import shutil
# =========== HYPERPARAMETERS ==========
MULTIVARIATE_DISTRIBUTIONS = ['m_gaussian_0_0', 'm_gaussian_1_1']
NUM_SAMPLES = 20000
NUM_TRIALS = 3
# ========== OUTPUT DIRECTORIES ==========
OUTPUT_DIR = 'MULTIVARIATE_OUTPUT/'
MODELS_OUTPUT_DIR = OUTPUT_DIR + 'MODELS/'
SYN_DATA_OUTPUT_DIR = OUTPUT_DIR + 'SYN_DATA/'
REAL_DATA_OUTPUT_DIR = OUTPUT_DIR + 'REAL_DATA/'
POWER_OUTPUT_DIR = OUTPUT_DIR + 'POWER/'
RESULTS_DIR = 'MULTIVARIATE_RESULTS/'
shutil.rmtree(OUTPUT_DIR, ignore_errors=True)
os.makedirs(MODELS_OUTPUT_DIR)
os.makedirs(SYN_DATA_OUTPUT_DIR)
os.makedirs(REAL_DATA_OUTPUT_DIR)
os.makedirs(POWER_OUTPUT_DIR)
os.makedirs(RESULTS_DIR)
# ========== RUN PIPELINE ==========
def generate_real_cmd(dist, num_samples, output_dir):
return 'python3 sample_prob_dist.py {0} {1} {2}/'.format(dist, num_samples, output_dir)
def train_gan_cmd(real_data_dir, output_dir):
return 'python3 train_prob_gan.py {0}data.npy {1}'.format(real_data_dir, output_dir)
def generate_syn_cmd(gen_dir, num_samples, output_dir):
return 'python3 generate_prob_gan.py {0}generator {1} {2}'.format(gen_dir, num_samples, output_dir)
def power_analysis_cmd(real_data_1_dir, real_data_2_dir, syn_data_1_dir, syn_data_2_dir, output_dir):
return 'python3 multivariate_power_analysis.py {0}data.npy {1}data.npy {2}data.npy {3}data.npy {4}'.format(real_data_1_dir, syn_data_1_dir, real_data_2_dir, syn_data_2_dir, output_dir)
def output_dirs(dist, k):
model_tag_base = '[{0}]_[k={1}]'.format(dist, k)
model_1_tag = model_tag_base + '_[v=1]'
model_2_tag = model_tag_base + '_[v=2]'
real_data_1_dir = '{0}{1}/'.format(REAL_DATA_OUTPUT_DIR, model_1_tag)
real_data_2_dir = '{0}{1}/'.format(REAL_DATA_OUTPUT_DIR, model_2_tag)
model_1_dir = '{0}{1}/'.format(MODELS_OUTPUT_DIR, model_1_tag)
model_2_dir = '{0}{1}/'.format(MODELS_OUTPUT_DIR, model_2_tag)
syn_data_1_dir = '{0}{1}/'.format(SYN_DATA_OUTPUT_DIR, model_1_tag)
syn_data_2_dir = '{0}{1}/'.format(SYN_DATA_OUTPUT_DIR, model_2_tag)
return real_data_1_dir, real_data_2_dir, model_1_dir, model_2_dir, syn_data_1_dir, syn_data_2_dir
def run_cmd_sequence(cmds):
for cmd in cmds:
os.system(cmd)
def generate_real_data_samples():
for i in range(len(MULTIVARIATE_DISTRIBUTIONS)):
for k in range(NUM_TRIALS):
dist_i = MULTIVARIATE_DISTRIBUTIONS[i]
real_data_1_dir, real_data_2_dir, _, _, _, _ = output_dirs(dist_i, k)
sample_real_1 = generate_real_cmd(dist_i, NUM_SAMPLES, real_data_1_dir)
sample_real_2 = generate_real_cmd(dist_i, NUM_SAMPLES, real_data_2_dir)
run_cmd_sequence([sample_real_1, sample_real_2])
def train_gans():
for i in range(len(MULTIVARIATE_DISTRIBUTIONS)):
for k in range(NUM_TRIALS):
dist_i = MULTIVARIATE_DISTRIBUTIONS[i]
real_data_1_dir, real_data_2_dir, model_1_dir, model_2_dir, _, _ = output_dirs(dist_i, k)
train_gan_1 = train_gan_cmd(real_data_1_dir, model_1_dir)
train_gan_2 = train_gan_cmd(real_data_2_dir, model_2_dir)
run_cmd_sequence([train_gan_1, train_gan_2])
def generate_syn_data_samples():
for i in range(len(MULTIVARIATE_DISTRIBUTIONS)):
for k in range(NUM_TRIALS):
dist_i = MULTIVARIATE_DISTRIBUTIONS[i]
_, _, model_1_dir, model_2_dir, syn_data_1_dir, syn_data_2_dir = output_dirs(dist_i, k)
sample_syn_1 = generate_syn_cmd(model_1_dir, NUM_SAMPLES, syn_data_1_dir)
sample_syn_2 = generate_syn_cmd(model_2_dir, NUM_SAMPLES, syn_data_2_dir)
run_cmd_sequence([sample_syn_1, sample_syn_2])
def run_power_analyses():
for i in range(len(MULTIVARIATE_DISTRIBUTIONS)):
for j in range(i, len(MULTIVARIATE_DISTRIBUTIONS)):
for k in range(NUM_TRIALS):
dist_i = MULTIVARIATE_DISTRIBUTIONS[i]
dist_j = MULTIVARIATE_DISTRIBUTIONS[j]
real_data_1_dir_i, real_data_2_dir_i, _, _, syn_data_1_dir_i, syn_data_2_dir_i = output_dirs(dist_i, k)
real_data_1_dir_j, real_data_2_dir_j, _, _, syn_data_1_dir_j, syn_data_2_dir_j = output_dirs(dist_j, k)
output_dir = '{0}[{1}_VS_{2}]_[k={3}]/'.format(POWER_OUTPUT_DIR, dist_i, dist_j, k)
cmd = power_analysis_cmd(real_data_1_dir_i, real_data_2_dir_j, syn_data_1_dir_i, syn_data_2_dir_j, output_dir)
run_cmd_sequence([cmd])
def visualize():
for i in range(len(MULTIVARIATE_DISTRIBUTIONS)):
for j in range(i, len(MULTIVARIATE_DISTRIBUTIONS)):
figure, axes = plt.subplots(nrows=2, ncols=1)
n = None
fdr_test_real_power = []
mmd_test_real_power = []
fdr_test_syn_power = []
mmd_test_syn_power = []
mmd_conservative_test_syn_power = []
for k in range(NUM_TRIALS):
dist_i = MULTIVARIATE_DISTRIBUTIONS[i]
dist_j = MULTIVARIATE_DISTRIBUTIONS[j]
power_dir_k = '{0}[{1}_VS_{2}]_[k={3}]/'.format(POWER_OUTPUT_DIR, dist_i, dist_j, k)
if n is None:
n = np.load(power_dir_k+'n.npy')
fdr_test_real_power.append(np.load(power_dir_k+'fdr_test_real_power.npy'))
mmd_test_real_power.append(np.load(power_dir_k+'mmd_test_real_power.npy'))
fdr_test_syn_power.append(np.load(power_dir_k+'fdr_test_syn_power.npy'))
mmd_test_syn_power.append(np.load(power_dir_k+'mmd_test_syn_power.npy'))
mmd_conservative_test_syn_power.append(np.load(power_dir_k+'mmd_conservative_test_syn_power.npy'))
n = np.array(n)
t_test_real_power = np.array(fdr_test_real_power)
mmd_test_real_power = np.array(mmd_test_real_power)
t_test_syn_power = np.array(fdr_test_syn_power)
mmd_test_syn_power = np.array(mmd_test_syn_power)
mmd_conservative_test_syn_power = np.array(mmd_conservative_test_syn_power)
# Plot curve of n vs power
sns.tsplot(data=fdr_test_real_power, time=n, ci=[68, 95], color='blue', condition='Real', ax=axes[0])
sns.tsplot(data=fdr_test_syn_power, time=n, ci=[68, 95], color='orange', condition='Synthetic', ax=axes[0])
axes[0].set_title('Sample Size vs FDR Corrected T Test Power')
axes[0].set_xlabel('Sample Size')
axes[0].set_ylabel('Power')
axes[0].set_ylim([-0.1, 1.1])
axes[0].legend(loc="upper right")
sns.tsplot(data=mmd_test_real_power, time=n, ci=[68, 95], color='blue', condition='Real', ax=axes[1])
sns.tsplot(data=mmd_test_syn_power, time=n, ci=[68, 95], color='orange', condition='Synthetic', ax=axes[1])
sns.tsplot(data=mmd_conservative_test_syn_power, time=n, ci=[68, 95], color='red', condition='Syn Conservative', ax=axes[1])
axes[1].set_title('Sample Size vs MMD Test Power')
axes[1].set_xlabel('Sample Size')
axes[1].set_ylabel('Power')
axes[1].set_ylim([-0.1, 1.1])
axes[1].legend(loc="upper right")
# Save results
figure.tight_layout()
figure.savefig('{0}{1}_VS_{2}.png'.format(RESULTS_DIR, dist_i, dist_j))
# ========== MAIN ==========
generate_real_data_samples()
train_gans()
generate_syn_data_samples()
run_power_analyses()
visualize()