-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathclassifier_temp.py
325 lines (277 loc) · 20.3 KB
/
classifier_temp.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
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
import csv
import random
import numpy as np
import voting
import demographic
import threading
from sklearn import svm
from sklearn.cluster import KMeans
from sklearn.svm import SVR
from sklearn import linear_model
counties = ["Alameda", "Butte" , "Contra Costa", "El Dorado", "Fresno",
"Humboldt", "Imperial", "Kern", "Kings", "Lake", "Los Angeles", "Madera",
"Marin", "Mendocino", "Merced", "Monterey", "Napa", "Nevada", "Orange",
"Placer", "Riverside", "Sacramento", "San Bernardino", "San Diego",
"San Francisco", "San Joaquin", "San Luis Obispo", "San Mateo",
"Santa Barbara", "Santa Clara", "Santa Cruz", "Shasta", "Solano", "Sonoma",
"Stanislaus", "Sutter", "Tulare", "Ventura", "Yolo", "Yuba"]
counties_dict = {}
for county in counties:
counties_dict[county] = 0
features = []
def compose_design_matrix(issue, tag):
designMatrix = None
targetMatrix = None
for county in counties:
feature_vec, target_vec = demographic.construct_submatrix(issue["year"], tag["demographics"], county, issue["prop"], tag["type"], issue["polarity"])
if designMatrix is None:
designMatrix = feature_vec
else:
designMatrix = np.vstack((designMatrix, feature_vec))
if targetMatrix is None:
targetMatrix = target_vec
else:
targetMatrix = np.vstack((targetMatrix, target_vec))
return designMatrix, targetMatrix
def combine_design_matrices(issues, tag):
designMatrix = None
targetMatrix = None
for issue in issues:
temp_design, temp_target = compose_design_matrix(issue, tag)
if designMatrix is None:
designMatrix = temp_design
else:
designMatrix = np.vstack((designMatrix, temp_design))
if targetMatrix is None:
targetMatrix = temp_target
else:
targetMatrix = np.vstack((targetMatrix, temp_target))
return designMatrix, targetMatrix
def zero_or_one(number):
if number > 50:
return 1
else:
return 0
def convert_to_binary_target(targetMatrix):
vecfunc = np.vectorize(zero_or_one)
return vecfunc(targetMatrix)
def is_number(s):
try:
float(s)
return True
except ValueError:
return False
def test():
sample_issues = [{ "year": 2008, "prop": "11", "polarity": "No" }]
sample_tag = { "name": "DiscoShit", "type": "Percent", "demographics": [26] }
model, design_matrix, target_matrix = build_classifier_model(sample_issues, sample_tag)
test_classifier_model(model, design_matrix, target_matrix.ravel())
def get_all_training_issues():
issues_hash = {}
issues_hash['infra'] = []
issues_hash['education'] = []
issues_hash['crime'] = []
issues_hash['environment'] = []
issues_hash['gambling'] = []
issues_hash['society'] = []
issues_hash['politics'] = []
issues_hash['corporate'] = []
f = open('Proposition Labels.csv', 'rU')
for line in f:
arr = line.split(',')
if arr[0] != '2014' and is_number(arr[0]):
next_dict = {'year': int(arr[0]), 'prop': arr[4], 'polarity': arr[6]}
category = arr[5]
if category != '0' and category != 'veterans':
issues_hash[category].append(next_dict)
return issues_hash
def build_classifier_model(issues, tag):
design_matrix, target_matrix = combine_design_matrices(issues, tag)
binary_target_matrix = convert_to_binary_target(target_matrix)
clf = svm.SVC(kernel="linear")
clf.fit(design_matrix, np.asarray(binary_target_matrix).ravel().transpose())
return clf, design_matrix, binary_target_matrix
def build_regression_model(issues, tag):
design_matrix, target_matrix = combine_design_matrices(issues, tag)
clf = SVR()
clf.fit(design_matrix, np.asarray(target_matrix).ravel().transpose())
clf.coeff_
return clf
def test_classifier_model(model, design_matrix, target_matrix, test_design_matrix, test_target_matrix, issue_tag, train_issues):
num_errors = 0.0
target_array = np.asarray(target_matrix).ravel()
for i in range(len(target_array) - 1):
if model.predict(design_matrix[i])[0] != np.array(target_array[i]):
num_errors += 1
num_test_errors = 0.0
test_target_array = np.asarray(test_target_matrix).ravel()
for i in range(len(test_target_array) - 1):
if model.predict(test_design_matrix[i])[0] != np.array(test_target_array[i]):
county = counties[i%40]
issue = issue_tag
prop = train_issues[i/40]
counties_dict[county] = counties_dict[county] + 1
print "County: %s \t Issue: %s \t Prop: %s \t Errors In County: %s" % (county, issue, prop, counties_dict[county])
num_test_errors += 1
print num_test_errors
return (num_errors / len(target_array)), (num_test_errors / len(test_target_array))
def train_model():
training_issues_hash = get_all_training_issues()
model_hash = {}
tag = { "name": "DiscoShit", "type": "Percent", "demographics": [26] }
def loop_test_features():
features = ["HC02_EST_VC129","HC02_EST_VC120", "HC02_EST_VC73", "HC02_EST_VC108", "HC02_EST_VC68"]
total_training = 0.0
total_test = 0.0
num_iters = 1
for i in range(num_iters):
train_error, test_error = test_features('politics', features)
total_training += train_error
total_test += test_error
def test_features(issue_tag, features):
training_issues_hash = get_all_training_issues()
train_issues = training_issues_hash[issue_tag]
test_size = len(train_issues) / 4
random.shuffle(train_issues)
test_issues = []
for i in range(test_size):
test_issues.append(train_issues.pop())
tag = { "name": "DiscoShit", "type": "Percent", "demographics": features }
model,design_matrix, target_matrix = build_classifier_model(train_issues, tag)
test_design_matrix,test_target_matrix = combine_design_matrices(test_issues, tag)
test_target_matrix = convert_to_binary_target(test_target_matrix)
train_error, test_error = test_classifier_model(model, design_matrix, target_matrix, test_design_matrix, test_target_matrix, issue_tag, train_issues)
return train_error, test_error
def load_features():
with open("features.txt", "rbU") as file:
contents = file.readline().strip()
features = contents.split(",")
return features
def brute_force_features():
all_features = load_features()
best = []
test_error = float("inf")
for i in range(20, 21):
for j in range(20):
features = random.sample(all_features, i)
avg = 0.0
k = 10
while k > 0:
try:
train, test = test_features('infra', features)
avg += test
k -= 1
except:
features = random.sample(all_features, i)
k = 10
avg /= 10.0
if avg < test_error:
best = features
test_error = avg
print "Best features:"
print best
print "Best test error:"
print test_error
def aggregate_test_features(tag, features):
N = 20
total_train = 0.0
total_test = 0.0
for i in range(N):
train, test = test_features(tag, features)
total_train += train
total_test += test
print "Training error:"
print total_train / N
print "Test error:"
print total_test / N
return test_classifier_model(model, design_matrix, target_matrix, test_design_matrix, test_target_matrix)
issues = ['infra','education','crime','environment','gambling','society','politics','corporate']
all_features = ["HC01_EST_VC03","HC01_MOE_VC03","HC02_EST_VC03","HC02_MOE_VC03","HC01_EST_VC04","HC01_MOE_VC04","HC02_EST_VC04","HC02_MOE_VC04","HC01_EST_VC05","HC01_MOE_VC05","HC02_EST_VC05","HC02_MOE_VC05","HC01_EST_VC06","HC01_MOE_VC06","HC02_EST_VC06","HC02_MOE_VC06","HC01_EST_VC07","HC01_MOE_VC07","HC02_EST_VC07","HC02_MOE_VC07","HC01_EST_VC08","HC01_MOE_VC08","HC02_EST_VC08","HC02_MOE_VC08","HC01_EST_VC09","HC01_MOE_VC09","HC02_EST_VC09","HC02_MOE_VC09","HC01_EST_VC10","HC01_MOE_VC10","HC02_EST_VC10","HC02_MOE_VC10","HC01_EST_VC11","HC01_MOE_VC11","HC02_EST_VC11","HC02_MOE_VC11","HC01_EST_VC12","HC01_MOE_VC12","HC02_EST_VC12","HC02_MOE_VC12","HC01_EST_VC13","HC01_MOE_VC13","HC02_EST_VC13","HC02_MOE_VC13","HC01_EST_VC14","HC01_MOE_VC14","HC02_EST_VC14","HC02_MOE_VC14","HC01_EST_VC15","HC01_MOE_VC15","HC02_EST_VC15","HC02_MOE_VC15","HC01_EST_VC16","HC01_MOE_VC16","HC02_EST_VC16","HC02_MOE_VC16","HC01_EST_VC17","HC01_MOE_VC17","HC02_EST_VC17","HC02_MOE_VC17","HC01_EST_VC19","HC01_MOE_VC19","HC02_EST_VC19","HC02_MOE_VC19","HC01_EST_VC20","HC01_MOE_VC20","HC02_EST_VC20","HC02_MOE_VC20","HC01_EST_VC21","HC01_MOE_VC21","HC02_EST_VC21","HC02_MOE_VC21","HC01_EST_VC22","HC01_MOE_VC22","HC02_EST_VC22","HC02_MOE_VC22","HC01_EST_VC23","HC01_MOE_VC23","HC02_EST_VC23","HC02_MOE_VC23","HC01_EST_VC24","HC01_MOE_VC24","HC02_EST_VC24","HC02_MOE_VC24","HC01_EST_VC25","HC01_MOE_VC25","HC02_EST_VC25","HC02_MOE_VC25","HC01_EST_VC27","HC01_MOE_VC27","HC02_EST_VC27","HC02_MOE_VC27","HC01_EST_VC28","HC01_MOE_VC28","HC02_EST_VC28","HC02_MOE_VC28","HC01_EST_VC29","HC01_MOE_VC29","HC02_EST_VC29","HC02_MOE_VC29","HC01_EST_VC30","HC01_MOE_VC30","HC02_EST_VC30","HC02_MOE_VC30","HC01_EST_VC31","HC01_MOE_VC31","HC02_EST_VC31","HC02_MOE_VC31","HC01_EST_VC32","HC01_MOE_VC32","HC02_EST_VC32","HC02_MOE_VC32","HC01_EST_VC34","HC01_MOE_VC34","HC02_EST_VC34","HC02_MOE_VC34","HC01_EST_VC35","HC01_MOE_VC35","HC02_EST_VC35","HC02_MOE_VC35","HC01_EST_VC36","HC01_MOE_VC36","HC02_EST_VC36","HC02_MOE_VC36","HC01_EST_VC37","HC01_MOE_VC37","HC02_EST_VC37","HC02_MOE_VC37","HC01_EST_VC38","HC01_MOE_VC38","HC02_EST_VC38","HC02_MOE_VC38","HC01_EST_VC40","HC01_MOE_VC40","HC02_EST_VC40","HC02_MOE_VC40","HC01_EST_VC41","HC01_MOE_VC41","HC02_EST_VC41","HC02_MOE_VC41","HC01_EST_VC42","HC01_MOE_VC42","HC02_EST_VC42","HC02_MOE_VC42","HC01_EST_VC43","HC01_MOE_VC43","HC02_EST_VC43","HC02_MOE_VC43","HC01_EST_VC44","HC01_MOE_VC44","HC02_EST_VC44","HC02_MOE_VC44","HC01_EST_VC45","HC01_MOE_VC45","HC02_EST_VC45","HC02_MOE_VC45","HC01_EST_VC46","HC01_MOE_VC46","HC02_EST_VC46","HC02_MOE_VC46", "HC01_EST_VC49","HC01_MOE_VC49","HC02_EST_VC49","HC02_MOE_VC49","HC01_EST_VC51","HC01_MOE_VC51","HC02_EST_VC51","HC02_MOE_VC51","HC01_EST_VC52","HC01_MOE_VC52","HC02_EST_VC52","HC02_MOE_VC52","HC01_EST_VC53","HC01_MOE_VC53","HC02_EST_VC53","HC02_MOE_VC53","HC01_EST_VC54","HC01_MOE_VC54","HC02_EST_VC54","HC02_MOE_VC54","HC01_EST_VC56","HC01_MOE_VC56","HC02_EST_VC56","HC02_MOE_VC56","HC01_EST_VC57","HC01_MOE_VC57","HC02_EST_VC57","HC02_MOE_VC57","HC01_EST_VC60","HC01_MOE_VC60","HC02_EST_VC60","HC02_MOE_VC60","HC01_EST_VC61","HC01_MOE_VC61","HC02_EST_VC61","HC02_MOE_VC61","HC01_EST_VC62","HC01_MOE_VC62","HC02_EST_VC62","HC02_MOE_VC62","HC01_EST_VC63","HC01_MOE_VC63","HC02_EST_VC63","HC02_MOE_VC63","HC01_EST_VC64","HC01_MOE_VC64","HC02_EST_VC64","HC02_MOE_VC64","HC01_EST_VC66","HC01_MOE_VC66","HC02_EST_VC66","HC02_MOE_VC66","HC01_EST_VC67","HC01_MOE_VC67","HC02_EST_VC67","HC02_MOE_VC67","HC01_EST_VC68","HC01_MOE_VC68","HC02_EST_VC68","HC02_MOE_VC68","HC01_EST_VC69","HC01_MOE_VC69","HC02_EST_VC69","HC02_MOE_VC69","HC01_EST_VC70","HC01_MOE_VC70","HC02_EST_VC70","HC02_MOE_VC70","HC01_EST_VC71","HC01_MOE_VC71","HC02_EST_VC71","HC02_MOE_VC71","HC01_EST_VC72","HC01_MOE_VC72","HC02_EST_VC72","HC02_MOE_VC72","HC01_EST_VC73","HC01_MOE_VC73","HC02_EST_VC73","HC02_MOE_VC73","HC01_EST_VC74","HC01_MOE_VC74","HC02_EST_VC74","HC02_MOE_VC74","HC01_EST_VC75","HC01_MOE_VC75","HC02_EST_VC75","HC02_MOE_VC75","HC01_EST_VC78","HC01_MOE_VC78","HC02_EST_VC78","HC02_MOE_VC78","HC01_EST_VC80","HC01_MOE_VC80","HC02_EST_VC80","HC02_MOE_VC80","HC01_EST_VC81","HC01_MOE_VC81","HC02_EST_VC81","HC02_MOE_VC81","HC01_EST_VC83","HC01_MOE_VC83","HC02_EST_VC83","HC02_MOE_VC83","HC01_EST_VC85","HC01_MOE_VC85","HC02_EST_VC85","HC02_MOE_VC85","HC01_EST_VC86","HC01_MOE_VC86","HC02_EST_VC86","HC02_MOE_VC86","HC01_EST_VC87","HC01_MOE_VC87","HC02_EST_VC87","HC02_MOE_VC87","HC01_EST_VC89","HC01_MOE_VC89","HC02_EST_VC89","HC02_MOE_VC89","HC01_EST_VC90","HC01_MOE_VC90","HC02_EST_VC90","HC02_MOE_VC90","HC01_EST_VC91","HC01_MOE_VC91","HC02_EST_VC91","HC02_MOE_VC91","HC01_EST_VC92","HC01_MOE_VC92","HC02_EST_VC92","HC02_MOE_VC92","HC01_EST_VC93","HC01_MOE_VC93","HC02_EST_VC93","HC02_MOE_VC93","HC01_EST_VC94","HC01_MOE_VC94","HC02_EST_VC94","HC02_MOE_VC94","HC01_EST_VC95","HC01_MOE_VC95","HC02_EST_VC95","HC02_MOE_VC95","HC01_EST_VC96","HC01_MOE_VC96","HC02_EST_VC96","HC02_MOE_VC96","HC01_EST_VC99","HC01_MOE_VC99","HC02_EST_VC99","HC02_MOE_VC99","HC01_EST_VC100","HC01_MOE_VC100","HC02_EST_VC100","HC02_MOE_VC100","HC01_EST_VC101","HC01_MOE_VC101","HC02_EST_VC101","HC02_MOE_VC101","HC01_EST_VC102","HC01_MOE_VC102","HC02_EST_VC102","HC02_MOE_VC102","HC01_EST_VC103","HC01_MOE_VC103","HC02_EST_VC103","HC02_MOE_VC103","HC01_EST_VC104","HC01_MOE_VC104","HC02_EST_VC104","HC02_MOE_VC104","HC01_EST_VC107","HC01_MOE_VC107","HC02_EST_VC107","HC02_MOE_VC107","HC01_EST_VC108","HC01_MOE_VC108","HC02_EST_VC108","HC02_MOE_VC108","HC01_EST_VC110","HC01_MOE_VC110","HC02_EST_VC110","HC02_MOE_VC110","HC01_EST_VC112","HC01_MOE_VC112","HC02_EST_VC112","HC02_MOE_VC112","HC01_EST_VC113","HC01_MOE_VC113","HC02_EST_VC113","HC02_MOE_VC113","HC01_EST_VC114","HC01_MOE_VC114","HC02_EST_VC114","HC02_MOE_VC114","HC01_EST_VC115","HC01_MOE_VC115","HC02_EST_VC115","HC02_MOE_VC115","HC01_EST_VC116","HC01_MOE_VC116","HC02_EST_VC116","HC02_MOE_VC116","HC01_EST_VC119","HC01_MOE_VC119","HC02_EST_VC119","HC02_MOE_VC119","HC01_EST_VC120","HC01_MOE_VC120","HC02_EST_VC120","HC02_MOE_VC120","HC01_EST_VC121","HC01_MOE_VC121","HC02_EST_VC121","HC02_MOE_VC121","HC01_EST_VC122","HC01_MOE_VC122","HC02_EST_VC122","HC02_MOE_VC122","HC01_EST_VC123","HC01_MOE_VC123","HC02_EST_VC123","HC02_MOE_VC123","HC01_EST_VC124","HC01_MOE_VC124","HC02_EST_VC124","HC02_MOE_VC124","HC01_EST_VC127","HC01_MOE_VC127","HC02_EST_VC127","HC02_MOE_VC127","HC01_EST_VC128","HC01_MOE_VC128","HC02_EST_VC128","HC02_MOE_VC128","HC01_EST_VC129","HC01_MOE_VC129","HC02_EST_VC129","HC02_MOE_VC129","HC01_EST_VC130","HC01_MOE_VC130","HC02_EST_VC130","HC02_MOE_VC130","HC01_EST_VC131","HC01_MOE_VC131","HC02_EST_VC131","HC02_MOE_VC131","HC01_EST_VC132","HC01_MOE_VC132","HC02_EST_VC132","HC02_MOE_VC132","HC01_EST_VC133","HC01_MOE_VC133","HC02_EST_VC133","HC02_MOE_VC133","HC01_EST_VC134","HC01_MOE_VC134","HC02_EST_VC134","HC02_MOE_VC134","HC01_EST_VC135","HC01_MOE_VC135","HC02_EST_VC135","HC02_MOE_VC135","HC01_EST_VC136","HC01_MOE_VC136","HC02_EST_VC136","HC02_MOE_VC136","HC01_EST_VC137","HC01_MOE_VC137","HC02_EST_VC137","HC02_MOE_VC137","HC01_EST_VC140","HC01_MOE_VC140","HC02_EST_VC140","HC02_MOE_VC140","HC01_EST_VC141","HC01_MOE_VC141","HC02_EST_VC141","HC02_MOE_VC141","HC01_EST_VC142","HC01_MOE_VC142","HC02_EST_VC142","HC02_MOE_VC142","HC01_EST_VC143","HC01_MOE_VC143","HC02_EST_VC143","HC02_MOE_VC143","HC01_EST_VC144","HC01_MOE_VC144","HC02_EST_VC144","HC02_MOE_VC144","HC01_EST_VC145","HC01_MOE_VC145","HC02_EST_VC145","HC02_MOE_VC145","HC01_EST_VC146","HC01_MOE_VC146","HC02_EST_VC146","HC02_MOE_VC146","HC01_EST_VC147","HC01_MOE_VC147","HC02_EST_VC147","HC02_MOE_VC147","HC01_EST_VC148","HC01_MOE_VC148","HC02_EST_VC148","HC02_MOE_VC148","HC01_EST_VC149","HC01_MOE_VC149","HC02_EST_VC149","HC02_MOE_VC149","HC01_EST_VC150","HC01_MOE_VC150","HC02_EST_VC150","HC02_MOE_VC150","HC01_EST_VC151","HC01_MOE_VC151","HC02_EST_VC151","HC02_MOE_VC151","HC01_EST_VC152","HC01_MOE_VC152","HC02_EST_VC152","HC02_MOE_VC152","HC01_EST_VC153","HC01_MOE_VC153","HC02_EST_VC153","HC02_MOE_VC153","HC01_EST_VC154","HC01_MOE_VC154","HC02_EST_VC154","HC02_MOE_VC154","HC01_EST_VC155","HC01_MOE_VC155","HC02_EST_VC155","HC02_MOE_VC155","HC01_EST_VC156","HC01_MOE_VC156","HC02_EST_VC156","HC02_MOE_VC156","HC01_EST_VC157","HC01_MOE_VC157","HC02_EST_VC157","HC02_MOE_VC157","HC01_EST_VC158","HC01_MOE_VC158","HC02_EST_VC158","HC02_MOE_VC158","HC01_EST_VC159","HC01_MOE_VC159","HC02_EST_VC159","HC02_MOE_VC159","HC01_EST_VC160","HC01_MOE_VC160","HC02_EST_VC160","HC02_MOE_VC160","HC01_EST_VC161","HC01_MOE_VC161","HC02_EST_VC161","HC02_MOE_VC161","HC01_EST_VC162","HC01_MOE_VC162","HC02_EST_VC162","HC02_MOE_VC162","HC01_EST_VC163","HC01_MOE_VC163","HC02_EST_VC163","HC02_MOE_VC163","HC01_EST_VC164","HC01_MOE_VC164","HC02_EST_VC164","HC02_MOE_VC164","HC01_EST_VC165","HC01_MOE_VC165","HC02_EST_VC165","HC02_MOE_VC165","HC01_EST_VC166","HC01_MOE_VC166","HC02_EST_VC166","HC02_MOE_VC166"]
crime_features = ['HC02_EST_VC32']#, 'HC02_EST_VC143', 'HC02_EST_VC32', 'HC02_EST_VC20', 'HC02_EST_VC08', 'HC02_EST_VC101', 'HC02_EST_VC122']
politics_features = ['HC02_EST_VC123', 'HC02_EST_VC108', 'HC02_EST_VC120', "HC02_EST_VC64", "HC02_EST_VC63", "HC02_EST_VC15", "HC02_EST_VC04"]
null_features = [
"HC02_EST_VC02",
"HC02_EST_VC19",
"HC02_EST_VC27",
"HC02_EST_VC40",
"HC02_EST_VC48",
"HC02_EST_VC59",
"HC02_EST_VC66",
"HC02_EST_VC77",
"HC02_EST_VC78",
"HC02_EST_VC80",
"HC02_EST_VC82",
"HC02_EST_VC84",
"HC02_EST_VC86",
"HC02_EST_VC89",
"HC02_EST_VC98",
"HC02_EST_VC106",
"HC02_EST_VC110",
"HC02_EST_VC111",
"HC02_EST_VC114",
"HC02_EST_VC118",
"HC02_EST_VC126",
"HC02_EST_VC139"
]
NUM_ITERS = 5
MAX_FEATURE_VECTOR_LENGTH = 10
def get_testing_error(tag, features, numIters):
testingError = 0
for i in range(numIters):
new_training_error, new_testing_error = test_features(tag, features)
testingError += new_testing_error
print "Tag: %s \t Features: %s \t TestingError: %s" % (tag, features, str(testingError/numIters))
return testingError/numIters
def find_best_features(tag, baseline_testing_error, features):
bestFeatures = {}
for feature in features.keys():
if feature[:8] == "HC02_EST":
try:
testing_error = get_testing_error(tag, [feature], NUM_ITERS)
new_testing_error = (testing_error + features[feature][0]*features[feature][1])/(features[feature][1] + 1)
new_count = features[feature][1] + 1
if new_testing_error < baseline_testing_error:
bestFeatures[feature] = [new_testing_error, new_count]
except ValueError, KeyError:
pass
print "Tag: %s \t Best Features Length: %s" % (tag, len(bestFeatures))
if len(bestFeatures) > MAX_FEATURE_VECTOR_LENGTH:
bestFeatures = find_best_features(tag, baseline_testing_error-.01, bestFeatures)
elif len(bestFeatures) == 0:
bestFeatures = ["HC02_EST_VC02"]
return bestFeatures
def tag_feature_selector(tag, baseline_error, features):
print tag
baseline_testing_error = baseline_error#get_testing_error(tag, ['HC02_EST_VC02'], NUM_BASE_ITERS)
print "Tag: %s \t Baseline Testing Error: %s" % (tag, baseline_testing_error)
best_features = find_best_features(tag, baseline_testing_error, features)
testing_error = get_testing_error(tag, best_features, NUM_ITERS*20)
print "Tag: %s \n Features: %s \n TestingError: %s \t Baseline Testing Error: %s" % (tag, best_features, testing_error, baseline_testing_error)
return baseline_testing_error, testing_error, best_features
class FeatureSelectorThread(threading.Thread):
def __init__(self,tag, baseline, features):
super(FeatureSelectorThread, self).__init__()
self.tag = tag
self.baseline_error = baseline
self.testing_error = 0
self.best_features = {}
for feature in features:
self.best_features[feature] = [0, 0]
def run(self):
self.baseline_error, self.testing_error, self.best_features = tag_feature_selector(self.tag, self.baseline_error, self.best_features)
def multithreaded_feature_finder():
threads = [FeatureSelectorThread('crime', .45, all_features), FeatureSelectorThread('education', .27, all_features),FeatureSelectorThread('infra', .25, all_features),FeatureSelectorThread('environment', .45, all_features),FeatureSelectorThread('society', .45, all_features),FeatureSelectorThread('politics', .42, all_features),FeatureSelectorThread('corporate', .35, all_features)]
for thread in threads:
thread.start()
for thread in threads:
thread.join()
for thread in threads:
print "Tag: %s \n Features: %s \n TestingError: %s \t Baseline Testing Error: %s" % (thread.tag, thread.best_features, thread.testing_error, thread.baseline_error)
def feature_evaluator(tag):
try:
testing_error = get_testing_error('crime', crime_features, NUM_ITERS * 4)
except ValueError, KeyError:
pass
print testing_error
multithreaded_feature_finder()
feature_evaluator()