-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathplot_revision.py
226 lines (187 loc) · 9.51 KB
/
plot_revision.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
"""
plot simulations results
"""
import numpy as np
import matplotlib.pyplot as plt
#import pickle as pkl
from config import SIMULATIONS_FIGURES_FOLDER
#from config import SIMULATIONS_ESTIMATED_FOLDER
from plot_functions import load_hierarchical_general_all
from plot_functions import load_enhanced_parallel_general_all
from plot_functions import plot_hierarchical, plot_enhanced_parallel
xticks_size = 14
yticks_size = 14
xlabel_size = 18
ylabel_size = 18
legend_size = 12
legend_loc = 'upper left'
linewidth = 3
linestyle = '-'
markersize = 5
marker = 'o'
#%%
'load every cumulative results we have'
idxs = [1,2,3,4,5]
## load results for main cases. LinSEM-TS-Gaussian
# hierarchical graphs
reg_2_1 = np.mean(load_hierarchical_general_all(2,1,idxs),0)
reg_2_2 = np.mean(load_hierarchical_general_all(2,2,idxs),0)
reg_2_3 = np.mean(load_hierarchical_general_all(2,3,idxs),0)
reg_3_1 = np.mean(load_hierarchical_general_all(3,1,idxs),0)
reg_3_2 = np.mean(load_hierarchical_general_all(3,2,idxs),0)
reg_3_3 = np.mean(load_hierarchical_general_all(3,3,idxs),0)
reg_4_1 = np.mean(load_hierarchical_general_all(4,1,idxs),0)
reg_4_2 = np.mean(load_hierarchical_general_all(4,2,idxs),0)
reg_4_3 = np.mean(load_hierarchical_general_all(4,3,idxs),0)
# enhanced parallel graphs
reg_5 = np.mean(load_enhanced_parallel_general_all(5,idxs),0)
reg_6 = np.mean(load_enhanced_parallel_general_all(6,idxs),0)
reg_7 = np.mean(load_enhanced_parallel_general_all(7,idxs),0)
reg_8 = np.mean(load_enhanced_parallel_general_all(8,idxs),0)
reg_9 = np.mean(load_enhanced_parallel_general_all(9,idxs),0)
## load results for UCB.
# hierarchical graphs
reg_ucb_2_1 = np.mean(load_hierarchical_general_all(2,1,idxs,ucb=True),0)
reg_ucb_2_2 = np.mean(load_hierarchical_general_all(2,2,idxs,ucb=True),0)
reg_ucb_2_3 = np.mean(load_hierarchical_general_all(2,3,idxs,ucb=True),0)
reg_ucb_3_1 = np.mean(load_hierarchical_general_all(3,1,idxs,ucb=True),0)
reg_ucb_3_2 = np.mean(load_hierarchical_general_all(3,2,idxs,ucb=True),0)
reg_ucb_3_3 = np.mean(load_hierarchical_general_all(3,3,idxs,ucb=True),0)
reg_ucb_4_1 = np.mean(load_hierarchical_general_all(4,1,idxs,ucb=True),0)
reg_ucb_4_2 = np.mean(load_hierarchical_general_all(4,2,idxs,ucb=True),0)
reg_ucb_4_3 = np.mean(load_hierarchical_general_all(4,3,idxs,ucb=True),0)
# enhanced parallel graphs
reg_ucb_5 = np.mean(load_enhanced_parallel_general_all(5,idxs,ucb=True),0)
reg_ucb_6 = np.mean(load_enhanced_parallel_general_all(6,idxs,ucb=True),0)
reg_ucb_7 = np.mean(load_enhanced_parallel_general_all(7,idxs,ucb=True),0)
reg_ucb_8 = np.mean(load_enhanced_parallel_general_all(8,idxs,ucb=True),0)
reg_ucb_9 = np.mean(load_enhanced_parallel_general_all(9,idxs,ucb=True),0)
## load results for known int. dist.
# hierarchical graphs
reg_known_dist_2_2 = np.mean(load_hierarchical_general_all(2,2,idxs,known_dist=True),0)
reg_known_dist_2_3 = np.mean(load_hierarchical_general_all(2,3,idxs,known_dist=True),0)
reg_known_dist_3_2 = np.mean(load_hierarchical_general_all(3,2,idxs,known_dist=True),0)
reg_known_dist_3_3 = np.mean(load_hierarchical_general_all(3,3,idxs,known_dist=True),0)
reg_known_dist_4_2 = np.mean(load_hierarchical_general_all(4,2,idxs,known_dist=True),0)
reg_known_dist_4_3 = np.mean(load_hierarchical_general_all(4,3,idxs,known_dist=True),0)
# graph misspecification. False positive edges. hieararchical
reg_mis_fp_2_2 = np.mean(load_hierarchical_general_all(2,2,idxs,mis_fp=True),0)
reg_mis_fp_2_3 = np.mean(load_hierarchical_general_all(2,3,idxs,mis_fp=True),0)
reg_mis_fp_3_2 = np.mean(load_hierarchical_general_all(3,2,idxs,mis_fp=True),0)
reg_mis_fp_3_3 = np.mean(load_hierarchical_general_all(3,3,idxs,mis_fp=True),0)
reg_mis_fp_4_2 = np.mean(load_hierarchical_general_all(4,2,idxs,mis_fp=True),0)
reg_mis_fp_4_3 = np.mean(load_hierarchical_general_all(4,3,idxs,mis_fp=True),0)
#%%
idxs = [1,2,3,4,5]
dL_pairs = [(2,1),(2,2),(2,3),(3,1),(3,2),(3,3),(4,1),(4,2),(4,3)]
N_range = [5,6,7,8,9]
'PLOT OUR GENERAL RESULTS ONLY'
for (d,L) in dL_pairs:
plot_hierarchical(d,L,idxs)
for N in N_range:
plot_enhanced_parallel(N, idxs)
'PLOT OUR GENERAL RESULTS ALONG WITH UCB'
for (d,L) in dL_pairs:
plot_hierarchical(d,L,idxs,compare_ucb=True)
for N in N_range:
plot_enhanced_parallel(N,idxs, compare_ucb=True)
'PLOT OUR KNOWN DIST RESULTS ONLY'
for (d,L) in dL_pairs:
plot_hierarchical(d,L,idxs,known_dist=True)
for N in N_range:
plot_enhanced_parallel(N,idxs,known_dist=True)
#%%
'FOR OUR HIERARCHICAL RESULTS, SET L CONSTANT, VARY d'
reg_L1 = [reg_2_1[-1],reg_3_1[-1],reg_4_1[-1]]
reg_L2 = [reg_2_2[-1],reg_3_2[-1],reg_4_2[-1]]
reg_L3 = [reg_2_3[-1],reg_3_3[-1],reg_4_3[-1]]
reg_ucb_L1 = [reg_ucb_2_1[-1],reg_ucb_3_1[-1],reg_ucb_4_1[-1]]
reg_ucb_L2 = [reg_ucb_2_2[-1],reg_ucb_3_2[-1],reg_ucb_4_2[-1]]
reg_ucb_L3 = [reg_ucb_2_3[-1],reg_ucb_3_3[-1],reg_ucb_4_3[-1]]
reg_known_dist_L1 = [reg_known_dist_2_1[-1],reg_known_dist_3_1[-1],reg_known_dist_4_1[-1]]
reg_known_dist_L2 = [reg_known_dist_2_2[-1],reg_known_dist_3_2[-1],reg_known_dist_4_2[-1]]
reg_known_dist_L3 = [reg_known_dist_2_3[-1],reg_known_dist_3_3[-1],reg_known_dist_4_3[-1]]
d_range = [2,3,4]
'plot for just our general case'
plt.figure()
plt.plot(d_range,reg_L1,label='L=1',markersize=10,linewidth=linewidth,linestyle='-',marker='o')
plt.plot(d_range,reg_L2,label='L=2',markersize=10,linewidth=linewidth,linestyle='-',marker='o')
plt.plot(d_range,reg_L3,label='L=3',markersize=10,linewidth=linewidth,linestyle='-',marker='o')
plt.xticks(d_range,fontsize=xticks_size)
plt.yticks(fontsize=yticks_size)
plt.xlabel('Degree $d$',size=xlabel_size)
plt.ylabel('Cumulative regret',size=ylabel_size)
plt.ylim(bottom=0)
plt.legend(fontsize=legend_size)
plt.tight_layout()
plt.grid()
plt.savefig(SIMULATIONS_FIGURES_FOLDER+'/hierarchical_vary_d.eps')
'compare with known dist'
plt.figure()
#plt.plot(d_range,reg_L1,'o',label='L=1',markersize=10,linewidth=linewidth,linestyle='-')
plt.plot(d_range,reg_L2,'o',label='L=2',markersize=10,linewidth=linewidth,linestyle='-')
plt.plot(d_range,reg_L3,'o',label='L=3',markersize=10,linewidth=linewidth,linestyle='-')
#plt.plot(d_range,reg_known_dist_L1,'o',label='L=1: known dist',markersize=10,linewidth=linewidth,linestyle='dashed')
plt.plot(d_range,reg_known_dist_L2,'o',label='L=2: known dist',markersize=10,linewidth=linewidth,linestyle='dashed')
plt.plot(d_range,reg_known_dist_L3,'o',label='L=3: known dist',markersize=10,linewidth=linewidth,linestyle='dashed')
plt.xticks(d_range,fontsize=xticks_size)
plt.yticks(fontsize=yticks_size)
plt.xlabel('Degree $d$',size=xlabel_size)
plt.ylabel('Cumulative regret',size=ylabel_size)
plt.ylim(bottom=0)
plt.legend(fontsize=legend_size)
plt.tight_layout()
plt.grid()
plt.savefig(SIMULATIONS_FIGURES_FOLDER+'/hierarchical_vary_d_compare_known_dist.eps')
#%%
'NOW DO IT FOR ENHANCED PARALLEL CASE'
N_range = [5,6,7,8,9]
reg_N = [reg_5[-1],reg_6[-1],reg_7[-1],reg_8[-1],reg_9[-1]]
reg_ucb_N = [reg_ucb_5[-1],reg_ucb_6[-1],reg_ucb_7[-1],reg_ucb_8[-1],reg_ucb_9[-1]]
reg_known_dist_N = [reg_known_dist_5[-1],reg_known_dist_6[-1],reg_known_dist_7[-1],reg_known_dist_8[-1],reg_known_dist_9[-1]]
plt.figure()
plt.plot(N_range,reg_N,'bo',label='LinSEM-TS-Gaussian',markersize=10,linewidth=linewidth,linestyle='-')
plt.plot(N_range,reg_ucb_N,'ro',label='UCB',markersize=10,linewidth=linewidth,linestyle='-')
#plt.plot(N_range,reg_known_dist_N,'ro',label='given dist. knowledge',markersize=10,linewidth=linewidth,linestyle='-')
plt.xticks(N_range,fontsize=xticks_size)
plt.yticks(fontsize=yticks_size)
plt.ylim(bottom=-50)
plt.xlabel('Graph size $N$',size=xlabel_size)
plt.ylabel('Cumulative regret',size=ylabel_size)
#plt.yscale('log')
#plt.ylim(bottom=0)
plt.legend(fontsize=legend_size)
plt.tight_layout()
plt.grid()
plt.savefig(SIMULATIONS_FIGURES_FOLDER+'/enhanced_vary_N_compare_ucb.eps')
#%%
'FINALLY SHOWCASE SOME GRAPH MISSPECIFICATION STUFF'
reg_mis_fp_L1 = [reg_mis_fp_2_1[-1],reg_mis_fp_3_1[-1],reg_mis_fp_4_1[-1]]
reg_mis_fp_L2 = [reg_mis_fp_2_2[-1],reg_mis_fp_3_2[-1],reg_mis_fp_4_2[-1]]
reg_mis_fp_L3 = [reg_mis_fp_2_3[-1],reg_mis_fp_3_3[-1],reg_mis_fp_4_3[-1]]
#%%
d_range = [2,3,4]
plt.figure()
#plt.plot(d_range,reg_L1,'o',label='L=1',markersize=10,linewidth=linewidth,linestyle='-')
plt.plot(d_range,reg_L2,'o',label='L=2',markersize=10,linewidth=linewidth,linestyle='-')
plt.plot(d_range,reg_L3,'o',label='L=3',markersize=10,linewidth=linewidth,linestyle='-')
#plt.plot(d_range,reg_mis_fp_L1,'o',label='L=1: graph misspec.',markersize=10,linewidth=linewidth,linestyle='dashed')
plt.plot(d_range,reg_mis_fp_L2,'o',label='L=2: graph misspec.',markersize=10,linewidth=linewidth,linestyle='dashed')
plt.plot(d_range,reg_mis_fp_L3,'o',label='L=3: graph misspec.',markersize=10,linewidth=linewidth,linestyle='dashed')
plt.xticks(d_range,fontsize=xticks_size)
plt.yticks(fontsize=yticks_size)
plt.xlabel('Degree $d$',size=xlabel_size)
plt.ylabel('Cumulative regret',size=ylabel_size)
plt.ylim(bottom=0)
plt.legend(fontsize=legend_size)
plt.tight_layout()
plt.grid()
plt.savefig(SIMULATIONS_FIGURES_FOLDER+'/hierarchical_vary_d_compare_mis_fp.eps')
#%%
plt.figure()
plt.plot(reg_2_2,label='d=2: grpah misspec.',markersize=markersize,linewidth=linewidth,linestyle='-')
plt.plot(reg_3_2,label='d=3: grpah misspec.',markersize=markersize,linewidth=linewidth,linestyle='-')
plt.plot(reg_4_2,label='d=4: grpah misspec.',markersize=markersize,linewidth=linewidth,linestyle='-')
plt.plot(reg_mis_fp_2_2,label='d=2: grpah misspec.',markersize=10,linewidth=linewidth,linestyle='-')
plt.plot(reg_mis_fp_3_2,label='d=3: graph misspec.',markersize=10,linewidth=linewidth,linestyle='-')
plt.plot(reg_mis_fp_4_2,label='d=4: graph misspec.',markersize=10,linewidth=linewidth,linestyle='-')