-
Notifications
You must be signed in to change notification settings - Fork 0
/
svm_pipe.py
41 lines (34 loc) · 1.03 KB
/
svm_pipe.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
#!/usr/bin/python
import pandas as pd
import sys
import pickle
import os
import pathlib
from numpy import loadtxt
import numpy as np
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.feature_extraction.text import TfidfTransformer
from sklearn.pipeline import Pipeline
from sklearn import svm
def training():
np.random.seed(100)
Corpus = pd.read_csv("Test.csv", encoding='utf-8-sig')
my_data = Corpus['text']
my_data1 = Corpus['label']
text_clf = Pipeline([
('vect', CountVectorizer()),
('tfidf', TfidfTransformer()),
('clf', svm.SVC(C=5, kernel='rbf', degree=3, gamma=0.5)),
])
#a = "Training...'\n"
my_clf = text_clf.fit(my_data, my_data1)
filename = 'Amz_5000_.sav'
pickle.dump(my_clf, open(filename, 'wb'))
abs=os.path.exists('Amz_5000_.sav')
lis = ["Successful : Training model Saved", "Oh Snap! something's wrong", str(len(Corpus))]
if abs is True:
del lis[1]
return lis
if abs is False:
del lis[0]
return lis