-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathplot.py
116 lines (97 loc) · 3.18 KB
/
plot.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
import numpy as np
from matplotlib import pyplot as plt
from lib.pcmc_utils import solve_ctmc
def plot(split,workFlag=True):
"""generates heatmap and error plots from locally saved data
to be run after infer.py
Double commented lines
Aguments:
split- proportion of data used for test set
"""
#set up for pyplot
if workFlag:
f,(ax1,ax2,ax3) = plt.subplots(1,3,figsize=(18,5))
else:
f,(ax1,ax2,ax3) = plt.subplots(1,3,figsize=(18,6))
plt.rcParams.update({'font.size': 12})
plt.rc('text', usetex=True)
plt.rc('font', family='serif')
params = {'text.latex.preamble' : [r'\usepackage{amsmath}']}
plt.rcParams.update(params)
#load errors
pcmc = np.load('pcmc_errors.npy')
mnl = np.load('mnl_errors.npy')
mmnl = np.load('mmnl_errors.npy')
#plotting variables
nep = pcmc.shape[1] #recover number of ways training data was split
percent = 100*(1-split)/nep*np.array(range(1,nep+1))
#make plots
ax1.plot(percent,np.mean(mnl,axis=0),marker='o',color='r',label='MNL')
ax1.plot(percent,np.mean(mmnl,axis=0),marker='o',color='g',label='MMNL')
ax1.plot(percent,np.mean(pcmc,axis=0),marker='o',color='b',label='PCMC')
#make it pretty
ax1.legend(loc=3)
ax1.set_xlabel(r'Percent of data used for training',labelpad=.1)
ax1.set_xticks(percent)
ax1.set_xlim((0,80))
ax1.set_ylabel(r'Error')
ax1.set_ylim(ymin=0)
#Now we plot the heatmap
#load params
Q= np.load('pcmc_params.npy')
n=Q.shape[0]
#sort the rows/columns by selection probabilities on whole set
d = solve_ctmc(Q)
idx = np.argsort(d)[::-1]
Q = Q[:,idx];Q = Q[idx,:]
#compute matrix C with c_ij=q_ij+q_ji (total rate for pairs)
#also scale P so that p_ij+p_ji=1
P = np.copy(Q)
C = np.zeros(P.shape)
C = C[:,idx];C = C[idx,:]
for i in range(n):
for j in range(i):
C[i,j]=P[i,j]+P[j,i]
C[j,i]=C[i,j]
P[i,j]/=C[i,j]
P[j,i]/=C[i,j]
#set meaningless diagonal values so that they will blend with colormap
for i in range(n):
C[i,i]=np.max(C)/2
P[i,i]=.5
#plot
cbar1=ax2.matshow(P,cmap=plt.cm.coolwarm)
if workFlag:
ax2.set_xticklabels(['','1','2','3','4','5','6'])#
ax2.set_yticklabels(['','1','2','3','4','5','6'])#
else:
ax2.set_xticklabels(['','1','2','3','4','5','6','7','8'])
ax2.set_yticklabels(['','1','2','3','4','5','6','7','8'])
ax2.tick_params(axis=u'both', which=u'both',length=0)
ax2.xaxis.tick_bottom()
cbar2=ax3.matshow(C,cmap=plt.cm.PuBuGn)
if workFlag:
ax3.set_xticklabels(['','1','2','3','4','5','6'])#
ax3.set_yticklabels(['','1','2','3','4','5','6'])#
else:
ax3.set_xticklabels(['','1','2','3','4','5','6','7','8'])
ax3.set_yticklabels(['','1','2','3','4','5','6','7','8'])
ax3.tick_params(axis=u'both', which=u'both',length=0)
ax3.xaxis.tick_bottom()
cbar1 = plt.colorbar(cbar1,ax=ax2)
cbar2 = plt.colorbar(cbar2,ax=ax3)
cbar1.set_label('$q_{ij}/(q_{ij}+q_{ji})$')
cbar2.set_label('$q_{ij}+q_{ji}$')
if workFlag:
ax2.set_title('SFwork')
else:
ax2.set_title('SFshop')
f.tight_layout(w_pad=0.2,h_pad=0)
if workFlag:
plt.savefig('workplot.png')
else:
plt.savefig('shopplot.png')
if __name__ == '__main__':
split=.25;#update this if you change it in infer.py
workFlag=True;#change to False to plot SFshop data
plot(split=split,workFlag=workFlag)