forked from RobeeF/M1DGMM
-
Notifications
You must be signed in to change notification settings - Fork 0
/
viz_test_on_simulated_data_partition_stability.py
238 lines (158 loc) · 7.91 KB
/
viz_test_on_simulated_data_partition_stability.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
# -*- coding: utf-8 -*-
"""
Created on Tue Jun 15 15:35:27 2021
@author: rfuchs
"""
import re
import os
os.chdir('C:/Users/rfuchs/Documents/GitHub/M1DGMM')
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.metrics.cluster import adjusted_rand_score
import autograd.numpy as np
results_path = 'C:/Users/rfuchs/Documents/These/Experiences/' # Results storage
os.chdir('C:/Users/rfuchs/Documents/These/Stats/mixed_dgmm/datasets')
datasets = os.listdir('simulated')
nb_trials = 30
nb_aris = int((nb_trials - 1) * (nb_trials / 2)) # Number of couples of ARIS to compute
n_clusters = 4
partition_repo = 'C:/Users/rfuchs/Documents/These/Experiences/similar_partition/'
designs = ['1n500', '1n1000', '1bisn500', '1bisn1000','2n500', '2n1000', '2bisn500', '2bisn1000']
###############################################################################
################################ Result analysis ############################
###############################################################################
#===========================================#
# MDGMM clustering
#===========================================#
# Small: r = {3, 2}, mini r = {2,1}, big = {5, 3, 1}
aris_distrib = pd.DataFrame()
archs = ['big', 'small', 'mini']
for arch in archs:
for dataset in datasets:
mdgmm_res = pd.read_csv(partition_repo + 'data/MDGMM/' + dataset[:-4] + '_' + arch + '.csv').values
assert mdgmm_res.shape[1] == 30
aris = [adjusted_rand_score(mdgmm_res[:,i], mdgmm_res[:,j]) \
for i in range(nb_trials) for j in range(nb_trials) if i < j]
aris_dataset = pd.DataFrame([aris, [dataset[6:-4]] * nb_aris, [arch] * nb_aris]).T
aris_dataset.columns = ['ARI_ij', 'design', 'Architecture']
aris_distrib = aris_distrib.append(aris_dataset,ignore_index = True)
ax = sns.boxplot(x = 'design', y = 'ARI_ij', hue = 'Architecture',
data=aris_distrib, palette="Set3", order = designs)
ax.tick_params('x', rotation = 30)
#ax.set_yscale("log")
ax.set_ylabel('Adjusted Rand Index')
#ax.set_title('ARI distribution over 30 runs for each data design')
plt.tight_layout()
#plt.savefig(results_path + 'similar_partition/figures/MDGMM_ARIs_distrib.png')
plt.show()
#===========================================#
# KMODES algorithm
#===========================================#
inits = ['Huang', 'Cao', 'random']
aris_distrib = pd.DataFrame()
for dataset in datasets:
for init in inits:
part_res_modes = pd.read_csv(partition_repo + 'data/KMODES/' + dataset[:-4] + '_' +\
init + '.csv').iloc[:,1:].values
assert part_res_modes.shape[1] == 30
aris = [adjusted_rand_score(part_res_modes[:,i], part_res_modes[:,j]) \
for i in range(nb_trials) for j in range(nb_trials) if i < j]
aris_dataset = pd.DataFrame([aris, [dataset[6:-4]] * nb_aris, [init] * nb_aris ]).T
aris_dataset.columns = ['ARI_ij', 'design', 'initialisation']
aris_distrib = aris_distrib.append(aris_dataset,ignore_index = True)
ax = sns.boxplot(x = 'design', hue = "initialisation", y = 'ARI_ij',
data=aris_distrib, palette="Set3", order = designs)
ax.legend(loc = 'lower right')
ax.tick_params('x', rotation = 30)
ax.set_ylabel('Adjusted Rand Index')
# Save the figs now..
plt.tight_layout()
plt.savefig(results_path + 'similar_partition/figures/KMODES_ARIs_distrib.png')
plt.show()
#===========================================#
# KPROTOTYPES
#===========================================#
inits = ['Huang', 'Cao', 'random']
aris_distrib = pd.DataFrame()
for dataset in datasets:
for init in inits:
part_res_proto = pd.read_csv(partition_repo + 'data/KPROTOTYPES/' + dataset[:-4] + '_' +\
init + '.csv').values
assert part_res_proto.shape[1] == 30
aris = [adjusted_rand_score(part_res_proto[:,i], part_res_proto[:,j]) \
for i in range(nb_trials) for j in range(nb_trials) if i < j]
aris_dataset = pd.DataFrame([aris, [dataset[6:-4]] * nb_aris, [init] * nb_aris ]).T
aris_dataset.columns = ['ARI_ij', 'design', 'initialisation']
aris_distrib = aris_distrib.append(aris_dataset,ignore_index = True)
ax = sns.boxplot(x = 'design', hue = "initialisation", y = 'ARI_ij',
data=aris_distrib, palette="Set3", order = designs)
ax.tick_params('x', rotation = 30)
ax.legend(loc = 'lower right')
ax.set_ylabel('Adjusted Rand Index')
# Save the figs now..
plt.tight_layout()
plt.savefig(results_path + 'similar_partition/figures/KPROTOTYPES_ARIs_distrib.png')
plt.show()
#===========================================#
# Hierarchical clustering
#===========================================#
# Perfectly deterministic
linkages = ['complete', 'average', 'single']
for design in designs:
for linky in linkages:
hierarch_res = pd.read_csv(results_path + 'similar_partition/data/Hierarchical/' +\
'result' + design + '_' + linky + '.csv', header = None)
assert hierarch_res.std(1).sum() == 0
#===========================================#
# SOM
#===========================================#
aris_distrib = pd.DataFrame()
sigmas = np.linspace(0.001, 3, 5)
lrs = np.linspace(0.0001, 0.5, 10)
for dataset in datasets:
for sig in sigmas:
for lr in lrs:
som_res = pd.read_csv(partition_repo + '/data/SOM/' + \
dataset[:-4] + '_' + str(sig) + '_' + str(lr) + '.csv').values
aris = [adjusted_rand_score(som_res[:,i], som_res[:,j]) \
for i in range(nb_trials) for j in range(nb_trials) if i < j]
aris_dataset = pd.DataFrame([aris, [dataset[6:-4]] * nb_aris,\
[str(sig)] * nb_aris, [str(lr)] * nb_aris ]).T
aris_dataset.columns = ['ARI_ij', 'design', 'sigma', 'learning rate']
aris_distrib = aris_distrib.append(aris_dataset,ignore_index = True)
fig, axs = plt.subplots(2, 4, figsize = (20, 10))
for d_idx, design in enumerate(designs):
data = aris_distrib[aris_distrib['design'] == design]
data['sigma'] = data['sigma'].astype(float).round(4)
data['learning rate'] = data['learning rate'].astype(float).round(3)
sns.boxplot(x = 'learning rate', hue = "sigma", y = 'ARI_ij',
data=data, palette="Set3", ax = axs[d_idx // 4, d_idx % 4])
if d_idx == len(designs) - 1:
axs[d_idx // 4, d_idx % 4].legend(title = 'sigma', markerscale=100.)
else:
axs[d_idx // 4, d_idx % 4].legend().set_visible(False)
axs[d_idx // 4, d_idx % 4].set_ylabel('ARI')
axs[d_idx // 4, d_idx % 4].set_xlabel('learning rate')
axs[d_idx // 4, d_idx % 4].set_title(design)
#plt.legend(sigmas, title = 'sigma', markerscale = 100.)
plt.tight_layout()
plt.savefig(results_path + 'similar_partition/figures/SOM_ARIs_distrib.png')
plt.show()
#===========================================#
# DBSCAN clustering
#===========================================#
lf_size = np.arange(1,6) * 10
epss = np.linspace(0.01, 5, 5)
min_ss = np.arange(1, 5)
data_to_fit = ['scaled', 'gower']
for design in designs:
for lfs in lf_size:
print("Leaf size:", lfs)
for eps in epss:
for min_s in min_ss:
for data in data_to_fit:
dbs_res= pd.read_csv(results_path + 'similar_partition/data/DBSCAN/dbscan' + \
dataset[:-4] + '_' + str(lfs) + '_' + str(eps) + '_' + str(min_s) + '_' +\
str(data) + '.csv', header = None)
assert dbs_res.std(1).sum() == 0