达观杯

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Fri Dec 28 00:10:58 2018

@author: yingjiezhang
"""


import pandas as pd 
import time 
from sklearn.svm import LinearSVC
from sklearn.linear_model import LogisticRegression
from sklearn.feature_extraction.text import TfidfVectorizer

#start = time.asctime(time.localtime(time.time()))
start = time.time()
print('start..................',start)
   
#如果内存不够可以尝试分批读进来，以下程序只是做个示范，先读1000行数据进来

df_train = pd.read_csv('train_set.csv')
df_test = pd.read_csv('test_set.csv')

#将文件划分为多个小文件，计算，再合并
'''

'''
df_train.drop(columns = ['article', 'id'], inplace = True)
df_test.drop(columns = ['article'], inplace = True)

#特征工程 将数据集中的字符文本转换为数字向量
vectorizer = TfidfVectorizer(ngram_range = (1,2), min_df = 3, max_df = 0.9, max_features = 100000)
#vectorizer.fit(df_train['word_seg'])
x_train = vectorizer.fit_transform(df_train['word_seg']) #fit_transform表示先使用fit方法，再transform方法
x_test = vectorizer.fit_transform(df_test['word_seg'])
y_train = df_train['class'] - 1    #之所以class-1是为了让label从0开始，好预测

#模型在这里设置为logistic回归，也可以改为其他的

lg = LogisticRegression(C=4, dual = True)
lg.fit(x_train, y_train)
y_test = lg.predict(x_test)


#将模型改为LinearSVC,将模型换了之后，为什么result没有变化？？
'''
lsvc = LinearSVC()
lsvc.fit(x_train,y_train)
y_test = lsvc.predict(x_test)
'''
df_test['class'] = y_test.tolist()
df_test['class'] = df_test['class'] + 1
df_result = df_test.loc[:, ['id', 'class']]
df_result.to_csv('result2.csv', index = False)

end = time.asctime(time.localtime(time.time()))
print('finished...............',end)

结果：

start… 1547092356.953075
finished… Thu Jan 10 12:03:10 2019

A榜得分：0.04

这个得分古怪的很！！
仔细检查，并没有发现什么不对，即使换过模型改过几遍之后，得分还是在0.04附近徘徊。