fileHandler_t.seek(0)
lines_t = fileHandler_t.readlines()
fileHandler_t.close()
fileHandler_tr = open('data\\data.txt')
fileHandler_tr.seek(0)
lines_tr = fileHandler_tr.readlines()
fileHandler_tr.close()
print "read finished"
vector_array = []
train_array = []
flag = []
for line in lines_t:
line = line.split('\t',1)
# cut word thrid arrary of sentence should change to 1 in test set
vector_array.append(vector_create(line[1], word_r, word_n))
pass
for line in lines_tr:
line = line.split('\t', 2)
train_array.append(vector_create(line[2], word_r, word_n))
flag.append(line[1])
pass
print "svm start"
# clf = svm.SVC()
clf = svm.LinearSVC(dual=False, C=10)
clf.fit(train_array, flag)
joblib.dump(clf, 'classifiter/svc_classifiter.pkl')
# clf = joblib.load('classifiter.pkl')
print "svm finished"
print "predicting"
fileHandler_out = open('dest_result/svc_result_14.csv', "w")
# print count
# print 'radio:', float(is_rubbish)/count
if(int(line[0])%5 == 0):
list_test.append(line)
else:
list_train.append(line)
pass
train_array=[]
test_array=[]
flag = []
answer = []
result = []
word_r, word_n = fileUtil.read_dict()
print 'set train vector'
for list1 in list_train:
train_array.append(vector_create(list1[2], word_r, word_n))
flag.append(list1[1])
pass
print 'set test vector'
for list2 in list_test:
test_array.append(vector_create(list2[2], word_r, word_n))
answer.append(list2[1])
pass
print "read finished"
# svm
clf = modle(train_array, flag)
result = predict(clf, test_array)
sum_rubbish = 0
sum_normal = 0
pre_rubbish = 0
pre_normal = 0
