-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathplot_functions.py
187 lines (144 loc) · 7.72 KB
/
plot_functions.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
"""
functions for quick plotting of the results
"""
import numpy as np
import matplotlib.pyplot as plt
import pickle as pkl
from config import SIMULATIONS_ESTIMATED_FOLDER, SIMULATIONS_FIGURES_FOLDER
xticks_size = 14
yticks_size = 14
xlabel_size = 18
ylabel_size = 18
legend_size = 12
legend_loc = 'upper left'
linewidth = 3
linestyle = '--'
markersize = 5
def load_hierarchical_general(d,L,idx,ucb=False,known_dist=False,known_noise=False,mis_fp=False,mis_fn=False):
# UCB is only run for general case: unknown dist, unknown noise
if ucb is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/hierarchical'+'/ucb_hierarchical_d_%d_l_%d_graph_%d.pkl'%(d,L,idx), 'rb'))
# known_dist result
elif known_dist is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/known_dist'+'/hierarchical_known_dist_d_%d_l_%d_graph_%d.pkl'%(d,L,idx), 'rb'))
# known_noise result
elif known_noise is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/known_noise'+'/hierarchical_known_noise_d_%d_l_%d_graph_%d.pkl'%(d,L,idx), 'rb'))
# graph misspec: false positive edges
elif mis_fp is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/misspecification_fp'+'/hierarchical_mis_fp_d_%d_l_%d_graph_%d.pkl'%(d,L,idx), 'rb'))
# graph misspec: false negative edges
elif mis_fn is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/misspecification_fn'+'/hierarchical_mis_fn_d_%d_l_%d_graph_%d.pkl'%(d,L,idx), 'rb'))
# if no special case is given, return the main results
else:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/hierarchical'+'/hierarchical_d_%d_l_%d_graph_%d.pkl'%(d,L,idx), 'rb'))
cum_reg = res['cum_reg']
avg_cum_reg = res['avg_cum_reg']
return avg_cum_reg, cum_reg, res
def load_hierarchical_general_all(d,L,indices,ucb=False,known_dist=False,known_noise=False,mis_fp=False,mis_fn=False):
all_avg_cum_reg = []
for idx in indices:
all_avg_cum_reg.append(load_hierarchical_general(d,L,idx,ucb,known_dist,known_noise,mis_fp,mis_fn)[0])
return np.asarray(all_avg_cum_reg)
def plot_hierarchical(d,L,indices,save=True,T=None,compare_ucb=False,known_dist=False,known_noise=False,mis_fp=False,mis_fn=False):
# load our algo's results. take average regret over graph instances
reg = np.mean(load_hierarchical_general_all(d,L,indices,False,known_dist,known_noise,mis_fp,mis_fn),0)
# if want to compare to ucb, also load that. take average regret over graph instances
if compare_ucb is True:
reg_ucb = np.mean(load_hierarchical_general_all(d,L,indices,True,known_dist,known_noise,mis_fp,mis_fn),0)
if compare_ucb is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/hierarchical_comp_d_%d_l_%d_avg.eps'%(d,L)
elif known_dist is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/hierarchical_known_dist_d_%d_l_%d_avg.eps'%(d,L)
elif known_noise is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/hierarchical_known_noise_d_%d_l_%d_avg.eps'%(d,L)
elif mis_fp is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/hierarchical_mis_fp_d_%d_l_%d_avg.eps'%(d,L)
elif mis_fn is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/hierarchical_mis_fn_d_%d_l_%d_avg.eps'%(d,L)
else:
save_name = SIMULATIONS_FIGURES_FOLDER+'/hierarchical_d_%d_l_%d_avg.eps'%(d,L)
if T is None:
if compare_ucb is True:
T = min(len(reg),len(reg_ucb))
else:
T = len(reg)
plt.figure()
# plot our results regardless
plt.plot(reg[:T],'b',markersize=markersize,label='LinSEM-TS-Gaussian',linewidth=linewidth,linestyle=linestyle)
if compare_ucb is True:
plt.plot(reg_ucb[:T],'r',markersize=markersize,label='UCB',linewidth=linewidth,linestyle=linestyle)
plt.xlabel('Number of Iterations',size=xlabel_size)
plt.ylabel('Cumulative regret',size=ylabel_size)
plt.xticks(fontsize=xticks_size)
plt.yticks(fontsize=yticks_size)
plt.legend(fontsize=legend_size,loc=legend_loc)
plt.tight_layout()
plt.grid()
if save is True:
plt.savefig(save_name)
def load_enhanced_parallel_general(N,idx,ucb=False,known_dist=False,known_noise=False,mis_fp=False,mis_fn=False):
# UCB is only run for general case: unknown dist, unknown noise
if ucb is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/enhanced_parallel'+'/ucb_size_N_%d_graph_%d.pkl'%(N,idx), 'rb'))
# known_dist result
elif known_dist is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/known_dist'+'/known_dist_size_N_%d_graph_%d.pkl'%(N,idx), 'rb'))
# known_noise result
elif known_noise is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/known_dist'+'/known_noise_size_N_%d_graph_%d.pkl'%(N,idx), 'rb'))
# graph misspec: false positive edges
elif mis_fp is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/misspecification_fp'+'/size_mis_fp_N_%d_graph_%d.pkl'%(N,idx), 'rb'))
# graph misspec: false negative edges
elif mis_fn is True:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/misspecification_fn'+'/size_mis_fn_N_%d_graph_%d.pkl'%(N,idx), 'rb'))
# if no special case is given, return the main results
else:
res = pkl.load(open(SIMULATIONS_ESTIMATED_FOLDER+'/enhanced_parallel'+'/size_N_%d_graph_%d.pkl'%(N,idx), 'rb'))
cum_reg = res['cum_reg']
avg_cum_reg = res['avg_cum_reg']
return avg_cum_reg, cum_reg, res
def load_enhanced_parallel_general_all(N,indices,ucb=False,known_dist=False,known_noise=False,mis_fp=False,mis_fn=False):
all_avg_cum_reg = []
for idx in indices:
all_avg_cum_reg.append(load_enhanced_parallel_general(N,idx,ucb,known_dist,known_noise,mis_fp,mis_fn)[0])
return np.asarray(all_avg_cum_reg)
def plot_enhanced_parallel(N,indices,save=True,T=None,compare_ucb=False,known_dist=False,known_noise=False,mis_fp=False,mis_fn=False):
# load our algo's results. take average regret over graph instances
reg = np.mean(load_enhanced_parallel_general_all(N,indices,False,known_dist,known_noise,mis_fp,mis_fn),0)
# if want to compare to ucb, also load that. take average regret over graph instances
if compare_ucb is True:
reg_ucb = np.mean(load_enhanced_parallel_general_all(N,indices,True,known_dist,known_noise,mis_fp,mis_fn),0)
if compare_ucb is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/enhanced_comp_size_N_%d.eps'%(N)
elif known_dist is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/enhanced_known_dist_size_N_%d.eps'%(N)
elif known_noise is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/enhanced_known_noise_size_N_%d.eps'%(N)
elif mis_fp is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/enhanced_mis_fp_size_N_%d.eps'%(N)
elif mis_fn is True:
save_name = SIMULATIONS_FIGURES_FOLDER+'/enhanced_mis_fn_size_N_%d.eps'%(N)
else:
save_name = SIMULATIONS_FIGURES_FOLDER+'/enhanced_size_N_%d.eps'%(N)
if T is None:
if compare_ucb is True:
T = min(len(reg),len(reg_ucb))
else:
T = len(reg)
plt.figure()
# plot our results regardless
plt.plot(reg[:T],'b',markersize=markersize,label='LinSEM-TS-Gaussian',linewidth=linewidth,linestyle=linestyle)
if compare_ucb is True:
plt.plot(reg_ucb[:T],'r',markersize=markersize,label='UCB',linewidth=linewidth,linestyle=linestyle)
plt.xlabel('Number of Iterations',size=xlabel_size)
plt.ylabel('Cumulative regret',size=ylabel_size)
plt.xticks(fontsize=xticks_size)
plt.yticks(fontsize=yticks_size)
plt.legend(fontsize=legend_size,loc=legend_loc)
plt.tight_layout()
plt.grid()
if save is True:
plt.savefig(save_name)