best_aro = sys.maxint
for i in range(50):
train_ids = ids
random.shuffle(train_ids)
all_ids = train_ids[141:]
train_ids = train_ids[0:140]
# calcultae valence and arousal find_a_v_mens
val_mean, aro_mean = find_a_v_mens_va(train_ids, valence, arousal)
train_feat = find_in_dict(feat, train_ids)
test_feat = find_in_dict(feat, all_ids)
# use regression
X_v, X_a = regression(train_feat, val_mean, aro_mean)
# calculating features for whole dataset
#print all_feat.shape
# use regresion function to calculate v and a
all_val = np.sum(np.array(test_feat) * X_v, axis=1)
all_aro = np.sum(test_feat * X_a, axis=1)
#print all_val.shape
#print all_aro.shape
print "ATTEMPT" + str(i)
avg = average_distance_va(all_val, all_aro, valence, arousal, all_ids)
nearest = nearest_dist_average_va(all_val, all_aro, valence, arousal,
Xtrain = X[1:trainlen]
Yvtrain = Yv[1:trainlen]
Yatrain = Ya[1:trainlen]
Xtest = X[trainlen + 1:]
Yvtest = Yv[trainlen + 1:]
Yatest = Ya[trainlen + 1:]
idstest = all_ids[trainlen + 1:]
print "check 5"
print X.shape
print len(Yv)
print len(Ya)
rv, ra = regression(Xtrain, Yvtrain, Yatrain)
Yvpred = np.sum(Xtest * rv, axis=1)
Yapred = np.sum(Xtest * ra, axis=1)
# print len(Yvpred)
# print len(Yvtest)
# print Yvpred.shape
# print Yvtest.shape
# avg = averagedist(Yvpred, Yapred, Yvtest, Yatest)
avg = average_distance_va(Yvpred, Yapred, valence, arousal, idstest)
if avg < best_avg:
best_avg = avg
print "BEST"
print 'Average distance: ' + str(best_avg)
for i in range(50):
train_ids = ids
random.shuffle(train_ids)
all_ids = train_ids[141:]
train_ids = train_ids[0:140]
# calcultae valence and arousal find_a_v_mens
val_mean, aro_mean = find_a_v_mens_va(train_ids, valence, arousal)
train_feat = find_in_dict(feat, train_ids)
test_feat = find_in_dict(feat, all_ids)
# use regression
X_v, X_a = regression(train_feat, val_mean, aro_mean)
# calculating features for whole dataset
#print all_feat.shape
# use regresion function to calculate v and a
all_val = np.sum(np.array(test_feat) * X_v, axis=1)
all_aro = np.sum(test_feat * X_a, axis=1)
#print all_val.shape
#print all_aro.shape
print "ATTEMPT" + str(i)
avg = average_distance_va(all_val, all_aro, valence, arousal, all_ids)
nearest = nearest_dist_average_va(all_val, all_aro, valence, arousal, all_ids)
Xtrain = X[1:trainlen]
Yvtrain = Yv[1:trainlen]
Yatrain = Ya[1:trainlen]
Xtest = X[trainlen + 1 :]
Yvtest = Yv[trainlen + 1 :]
Yatest = Ya[trainlen + 1 :]
idstest = all_ids[trainlen + 1 :]
print "check 5"
print X.shape
print len(Yv)
print len(Ya)
rv, ra = regression(Xtrain, Yvtrain, Yatrain)
Yvpred = np.sum(Xtest * rv, axis=1)
Yapred = np.sum(Xtest * ra, axis=1)
# print len(Yvpred)
# print len(Yvtest)
# print Yvpred.shape
# print Yvtest.shape
# avg = averagedist(Yvpred, Yapred, Yvtest, Yatest)
avg = average_distance_va(Yvpred, Yapred, valence, arousal, idstest)
if avg < best_avg:
best_avg = avg
print "BEST"
print "Average distance: " + str(best_avg)