forked from jackrgoetz/Mondrian_Tree_AL
-
Notifications
You must be signed in to change notification settings - Fork 0
/
make_graphs_het.py
118 lines (84 loc) · 4.43 KB
/
make_graphs_het.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
import matplotlib
matplotlib.use('AGG')
from collections import defaultdict
import matplotlib.pyplot as plt
import numpy as np
import glob
def main():
globby = glob.glob('graphs/het_mt*.npz')
print('{} data points'.format(len(globby)))
n_finals = [1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000]
data = None
for file in globby:
temp_data = np.load(file)
if data is None:
data = dict(temp_data)
else:
for key in temp_data:
data[key] += temp_data[key]
MT_al_MSE = data['MT_al_MSE'] / len(globby)
MT_rn_MSE = data['MT_rn_MSE'] / len(globby)
MT_uc_MSE = data['MT_uc_MSE'] / len(globby)
BT_al_MSE = data['BT_al_MSE'] / len(globby)
BT_rn_MSE = data['BT_rn_MSE'] / len(globby)
BT_uc_MSE = data['BT_uc_MSE'] / len(globby)
f, axarr = plt.subplots(2, sharex=True)
mt_al = axarr[0].plot(n_finals, MT_al_MSE, color = 'red', label='Mondrian Tree - Active sampling')
mt_rn = axarr[0].plot(n_finals, MT_rn_MSE, color = 'blue', label = 'Mondrian Tree - Random sampling')
mt_uc = axarr[0].plot(n_finals, MT_uc_MSE, color = 'green', label = 'Mondrian Tree - Uncertainty sampling')
axarr[0].set_title('Heteroskedastic sim, n={} trials'.format(len(globby)))
axarr[0].legend(loc='best')
bt_al = axarr[1].plot(n_finals, BT_al_MSE, color = 'red', linestyle = '--',
label = 'Breiman Tree - Active sampling')
bt_rn = axarr[1].plot(n_finals, BT_rn_MSE, color = 'blue', linestyle = '--',
label = 'Breiman Tree - Random sampling')
bt_rn = axarr[1].plot(n_finals, BT_uc_MSE, color = 'green', linestyle = '--',
label = 'Breiman Tree - Uncertainty sampling')
axarr[1].legend(loc='best')
f.text(0.01, 0.5, 'MSE', va='center', rotation='vertical')
f.text(0.5, 0.01, 'Final number of labelled points', ha='center')
variance_data = {
'MT_al_MSE': defaultdict(list),
'MT_rn_MSE': defaultdict(list),
'MT_uc_MSE': defaultdict(list),
'BT_al_MSE': defaultdict(list),
'BT_rn_MSE': defaultdict(list),
'BT_uc_MSE': defaultdict(list),
}
for file in globby:
temp_data = np.load(file)
for key in temp_data:
curr = temp_data[key]
for i in range(len(curr)):
variance_data[key][i].append(curr[i])
for key in variance_data:
for nidx in variance_data[key]:
variance_data[key][nidx] = np.array(variance_data[key][nidx])
MT_al_MSE_var = np.std(np.array(list(variance_data['MT_al_MSE'].values())), axis=1)
MT_rn_MSE_var = np.std(np.array(list(variance_data['MT_rn_MSE'].values())), axis=1)
MT_uc_MSE_var = np.std(np.array(list(variance_data['MT_uc_MSE'].values())), axis=1)
BT_al_MSE_var = np.std(np.array(list(variance_data['BT_al_MSE'].values())), axis=1)
BT_rn_MSE_var = np.std(np.array(list(variance_data['BT_rn_MSE'].values())), axis=1)
BT_uc_MSE_var = np.std(np.array(list(variance_data['BT_uc_MSE'].values())), axis=1)
mt_al_err = axarr[0].errorbar(n_finals, MT_al_MSE, MT_al_MSE_var, color = 'red', marker='^', capsize=10)
mt_rn_err = axarr[0].errorbar(n_finals, MT_rn_MSE, MT_rn_MSE_var, color = 'blue', marker='^', capsize=10)
mt_uc_err = axarr[0].errorbar(n_finals, MT_uc_MSE, MT_uc_MSE_var, color = 'green', marker='^', capsize=10)
bt_al_err = axarr[1].errorbar(n_finals, BT_al_MSE, BT_al_MSE_var, color = 'red', marker='^', capsize=10)
bt_rn_err = axarr[1].errorbar(n_finals, BT_rn_MSE, BT_rn_MSE_var, color = 'blue', marker='^', capsize=10)
bt_rn_err = axarr[1].errorbar(n_finals, BT_uc_MSE, BT_uc_MSE_var, color = 'green', marker='^', capsize=10)
plt.tight_layout()
plt.savefig('het_mt.png')
corrected_mt_al_vals = np.array(list(variance_data['MT_al_MSE'].values())) - np.array(list(variance_data['MT_rn_MSE'].values()))
corrected_bt_al_vals = np.array(list(variance_data['BT_al_MSE'].values())) - np.array(list(variance_data['BT_rn_MSE'].values()))
plt.figure()
plt.title("heteroskedastic Mondrian Trees boxplot normed MSE")
plt.boxplot(corrected_mt_al_vals.T, labels=n_finals)
plt.axhline(linewidth=1, color='r')
plt.savefig('het_corrected_boxplot_mt.png')
plt.figure()
plt.title("heteroskedastic Breiman Trees boxplot normed MSE")
plt.boxplot(corrected_bt_al_vals.T, labels=n_finals)
plt.axhline(linewidth=1, color='r')
plt.savefig('het_corrected_boxplot_bt.png')
if __name__ == '__main__':
main()