jinlonghe / textclassify Goto Github PK

data
├── test
│   ├── A_宾馆饭店
│   │   ├── bj6_seg_pos.txt
│   │   ├── 三亚市春节宾馆房价不乱涨价违者将受到严处_seg_pos.txt
│   │   └── 住宿-宾馆名录_seg_pos.txt
│   ├── B_城市概况
│   │   ├── bozhou02_seg_pos.txt
│   │   ├── yangzhou01_seg_pos.txt
│   │   └── zhaoqing04_seg_pos.txt
│   ├── C_地方文化
│   .........
│   .........
└── train
    ├── A_宾馆饭店
    │   ├── bj1.txt
    │   ├── 魏宝山景区.txt
    │   └── 龙 潭 瀑 布.txt
    │   .........
    │   .........
    └── H_休闲娱乐
        ├── banna01.txt
        └── 金牌、银牌表示推荐的娱乐场所。如果想了解娱乐场所的详细信息，请点击娱乐场所名称。.txt

"""
Created on Mon Dec  7 20:36:00 2015

@author: hehe
"""
import os
import numpy as np
from sklearn import linear_model

from TextClassify import BagOfWords
from TextClassify import TextClassify

data_dir = 'data'
## BAG OF WORDS MODEL
BOW = BagOfWords(os.path.join(data_dir, 'train'))

# 创建词典并且保存，如果保存过词典，以后直接load就行
BOW.build_dictionary()
BOW.save_dictionary(os.path.join(data_dir, 'dicitionary.pkl'))

# BOW.load_dictionary('dicitionary.pkl')

## LOAD DATA
train_feature, train_target = BOW.transform_data(os.path.join(data_dir, 'train'))
test_feature, test_target = BOW.transform_data(os.path.join(data_dir, 'test'))

## TRAIN LR MODEL
logreg = linear_model.LogisticRegression(C=1e5)
logreg.fit(train_feature, train_target)

## PREDICT
test_predict = logreg.predict(test_feature)

## ACCURACY

true_false = (test_predict == test_target)
accuracy = np.count_nonzero(true_false) / float(len(test_target))
print("accuracy is %f" % accuracy)

## TextClassify
TextClassifier = TextClassify()
pred = TextClassifier.text_classify('test.txt', BOW, logreg)
print(pred[0])