-
Notifications
You must be signed in to change notification settings - Fork 1
/
step5_linear.py
181 lines (146 loc) · 8.2 KB
/
step5_linear.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
#!/usr/bin/env python
# -*- coding: utf-8 -*-
""" This script uses the file produced in step 4 to produce the final figures
with model reports using linearSVR.
This program is free software: you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation, either version 3 of the License, or (at your option) any later
version.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
this program. If not, see <http://www.gnu.org/licenses/>.
"""
__authors__ = ["Zafeirios Fountas", "Kyriacos Nikiforou", "Anastasia Sylaidi"]
__credits__ = ["Warrick Roseboom", "Anil Seth", "Murray Shanahan"]
__license__ = "GPLv3"
__version__ = "0.1"
__maintainer__ = "Zafeirios Fountas"
__email__ = "fountas@outlook.com"
__status__ = "Published"
import matplotlib.pyplot as plt
import numpy as np
import pickle,csv
all_experiment_types = ['prosp_low', 'prosp_high', 'retro_low', 'retro_high']
print('---- LOADING VIDEO TRIAL SEQUENCE ----')
video_scene_types = {}
for i in [1,2,3,4,5,16,17,18, 19]: video_scene_types[i] = 'city'
for i in [6,7,8,9,10,11,12,13]: video_scene_types[i] = 'campus_outside'
for i in [14,15]: video_scene_types[i] = 'office_cafe'
for i in range(20,34): video_scene_types[i] = 'office_cafe'
video_trial_seq_keys = ['Trial','Video','Test_duration_secs','Test_duration_frames','Start_frame','End_frame']
video_trial_seq_temp = csv.reader(open('Video_trial_sequence.csv'))
video_trial_seq_temp = [row for row in video_trial_seq_temp]
if video_trial_seq_keys != video_trial_seq_temp[0]: exit()
else: video_trial_seq_temp = video_trial_seq_temp[1:]
video_ids = {}
scene_type_per_trial = {}
video_durations = {}
for i in range(len(video_trial_seq_temp)):
video_ids[int(float(video_trial_seq_temp[i][0]))] = int(float(video_trial_seq_temp[i][1])) # Just for making sure this works..!
scene_type_per_trial[int(float(video_trial_seq_temp[i][0]))] = video_scene_types[int(float(video_trial_seq_temp[i][1]))]
video_durations[int(float(video_trial_seq_temp[i][0]))] = float(video_trial_seq_temp[i][2]) # Just for double-checking..!
all_scene_types = ['city', 'campus_outside', 'office_cafe']
with open("model_accumulators_dataset.pkl", "rb") as file:
data = pickle.load(file)
print([x + ': ' + str(np.shape(np.array(data[x]))) for x in data.keys()])
from sklearn.neural_network import MLPRegressor
from sklearn.svm import SVR, LinearSVR
from sklearn import linear_model
from scipy import stats
#Y = np.array(data['durations'])
XX = {'prosp_low' :np.array(data['prosp_low']).transpose(),
'prosp_high':np.array(data['prosp_high']).transpose(),
'retro_low' :np.array(data['retro_low']).transpose(),
'retro_high':np.array(data['retro_high']).transpose()}
YY = {'prosp_low' :np.array(data['durations']['prosp_low']).transpose(),
'prosp_high':np.array(data['durations']['prosp_high']).transpose(),
'retro_low' :np.array(data['durations']['retro_low']).transpose(),
'retro_high':np.array(data['durations']['retro_high']).transpose()}
TRIALS = {'prosp_low' :np.array(data['trial_no']['prosp_low']).transpose(),
'prosp_high':np.array(data['trial_no']['prosp_high']).transpose(),
'retro_low' :np.array(data['trial_no']['retro_low']).transpose(),
'retro_high':np.array(data['trial_no']['retro_high']).transpose()}
Xtrain = {'city':[], 'office_cafe':[], 'campus_outside':[]}
Ytrain = {'city':[], 'office_cafe':[], 'campus_outside':[]}
minimum_length = min([len(YY[x]) for x in ['prosp_low', 'prosp_high']])
for i,mode in enumerate(['prosp_low', 'prosp_high']):
for i in range(minimum_length): #len(XX[mode])):
trial = TRIALS[mode][i]
scene_type = scene_type_per_trial[trial]
Xtrain[scene_type].append(XX[mode][i])
Ytrain[scene_type].append(YY[mode][i])
xx = np.array([1.0,1.5,2.0,3.0, 4.0, 6.0, 8.0, 12.0, 16.0, 24.0, 32.0, 48.0, 64.0])
while True:
SEED = np.random.randint(1,100)
SEED = 27 #awesome!
np.random.seed(SEED)
alg_all = {'city':LinearSVR(), 'office_cafe':LinearSVR(), 'campus_outside':LinearSVR()}
for scene_type in all_scene_types:
alg_all[scene_type].fit(Xtrain[scene_type],Ytrain[scene_type])
x_overall = {'prosp_low':[], 'prosp_high':[], 'retro_low':[], 'retro_high':[]}
y_overall = {'prosp_low':[], 'prosp_high':[], 'retro_low':[], 'retro_high':[]}
e_overall = {'prosp_low':[], 'prosp_high':[], 'retro_low':[], 'retro_high':[]}
for mode in ['prosp_low', 'prosp_high', 'retro_low', 'retro_high']:
for scene_type in all_scene_types:
X = []
Y = []
for x in range(len(XX[mode])):
trial = TRIALS[mode][x]
if scene_type == scene_type_per_trial[trial]:
X.append(XX[mode][x])
Y.append(YY[mode][x])
X = np.array(X)
Y = np.array(Y)
Yest = alg_all[scene_type].predict(X)
for y in range(len(Yest)):
x_overall[mode].append(Y[y])
y_overall[mode].append(Yest[y])
e_overall[mode].append((Yest[y]-Y[y])/Y[y])
# Now x,y,e_overall have the results for all scene_types!
# So scene doesn't need to be used from now on..
plt.figure(figsize=(16,6))
for m,mode in enumerate(['prosp_low', 'prosp_high', 'retro_low', 'retro_high']):
Y_binned = {}
for y in range(len(x_overall[mode])):
if x_overall[mode][y] not in Y_binned:
Y_binned[x_overall[mode][y]] = []
Y_binned[x_overall[mode][y]].append(y_overall[mode][y])
y_mean = np.array([np.mean(Y_binned[k]) for k in xx])
y_med = np.array([np.median(Y_binned[k]) for k in xx])
y_std = np.array([np.std(Y_binned[k]) for k in xx]) #y_sem = np.array([stats.sem(Y_binned[k]) for k in xx])
print('We ended up having:',sum([len(Y_binned[x]) for x in xx]),'datapoints for', mode)
plt.subplot(2,5,1+m)
plt.scatter(x_overall[mode][:],y_overall[mode],c='gray')
plt.plot(xx,y_mean,'k')
plt.plot(xx,y_med,'r')
plt.plot(xx,y_mean+y_std,'gray')
plt.plot(xx,y_mean-y_std,'gray')
plt.xlim(1,65)
plt.ylim(1,65)
plt.plot([1,64],[1,64],'k--')
plt.xscale('log')
plt.yscale('log')
plt.xlabel(mode+' ')
plt.subplot(2,5,5)
plt.plot([1.10, 1.10],[np.mean(e_overall['retro_low'])-stats.sem(e_overall['retro_low']), np.mean(e_overall['retro_low'])+stats.sem(e_overall['retro_low'])],c='k',lw=1.0)
plt.plot([2.10, 2.10],[np.mean(e_overall['retro_high'])-stats.sem(e_overall['retro_high']), np.mean(e_overall['retro_high'])+stats.sem(e_overall['retro_high'])],c='k',lw=1.0)
plt.plot([1.10,2.10], [np.mean(e_overall['retro_low']), np.mean(e_overall['retro_high'])],'k--',label='Retrospective')
plt.scatter([1.10,2.10], [np.mean(e_overall['retro_low']), np.mean(e_overall['retro_high'])],c='k')
plt.plot([1, 1],[np.mean(e_overall['prosp_low'])-stats.sem(e_overall['prosp_low']), np.mean(e_overall['prosp_low'])+stats.sem(e_overall['prosp_low'])],c='k',lw=1.0)
plt.plot([2, 2],[np.mean(e_overall['prosp_high'])-stats.sem(e_overall['prosp_high']), np.mean(e_overall['prosp_high'])+stats.sem(e_overall['prosp_high'])],c='k',lw=1.0)
plt.plot([1,2], [np.mean(e_overall['prosp_low']), np.mean(e_overall['prosp_high'])],'k',label='Prospective')
plt.scatter([1,2], [np.mean(e_overall['prosp_low']), np.mean(e_overall['prosp_high'])],c='k')
plt.xlim(0.5,2.5)
plt.xticks([],[])
print('SEED',SEED)
for scene_type in all_scene_types:
print('Parameters for', scene_type, alg_all[scene_type].get_params(),'weights:',alg_all[scene_type].coef_)
print('OK')
plt.tight_layout()
plt.savefig('figures/linear_1_only_pro_tr.png')
plt.savefig('figures/linear_1_only_pro_tr.svg')
plt.show()
exit()
#