任务
构建逻辑回归模型进行预测(在构建部分数据需要进行缺失值处理和数据类型转换,如果不能处理,可以直接暴力删除)
数据集
主要问题是根据数据建立一个逻辑回归模型来预测贷款是否逾期。
遇到的问题
-
encoding=‘gb18030’,为什么改为utf-8不可以?
-
读取数据时报错
data = pd.read_csv(path + ‘data.csv’)
‘utf-8’ codec can’t decode byte 0xbf in position 0: invalid start byte
解决办法data = pd.read_csv(path + ‘data.csv’,encoding=‘gb18030’)
-
对"latest_query_time"与"loans_latest_time"这两列怎么处理?
实现代码
主要分为五步:
1.数据处理
- 对缺失值进行处理
- 对包含中文数据进行映射
- 对两个时间的不确定进行删除
2.划分训练集和验证集,验证集比例为test_size
3. 提取特征和标签
4.模型训练与模型预测、保存模型
5.模型评分
#!/usr/bin/env python 3.6
#-*- coding:utf-8 -*-
# @File : V1.py
# @Date : 2018-11-14
# @Author : 黑桃
# @整理 : 等到的过去
import pickle
import pandas as pd #数据分析
from pandas import Series,DataFrame
from sklearn.model_selection import train_test_split
import time
from sklearn.externals import joblib
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import f1_score
print("开始......")
t_start = time.time()
path = "E:/MyPython/Machine_learning_GoGoGo/Task1/data_set/"
"""==========================================================================================================
1 读取数据
"""
print("数据预处理")
data = pd.read_csv(path + 'data.csv',encoding='gb18030')
"""
1.1 缺失值用100填充
"""
data=DataFrame(data.fillna(100))
"""
1.2 对reg_preference_for_trad 的处理 【映射】
nan=0 境外=1 一线=5 二线=2 三线 =3 其他=4
"""
n=set(data['reg_preference_for_trad'])
dic={}
for i,j in enumerate(n):
dic[j]=i
data['reg_preference_for_trad'] = data['reg_preference_for_trad'].map(dic)
"""
1.3 对source 的处理 【映射】
"""
n=set(data['source'])
dic={}
for i,j in enumerate(n):
dic[j]=i
data['source'] = data['source'].map(dic)
"""
1.4 对bank_card_no 的处理 【映射】
"""
n=set(data['bank_card_no'])
dic={}
for i,j in enumerate(n):
dic[j]=i
data['bank_card_no'] = data['bank_card_no'].map(dic)
"""
1.5 对 id_name的处理 【映射】
"""
n=set(data['id_name'])
dic={}
for i,j in enumerate(n):
dic[j]=i
data['id_name'] = data['id_name'].map(dic)
"""
1.6 对 time 的处理 【删除】
"""
data.drop(["latest_query_time"],axis=1,inplace=True)
data.drop(["loans_latest_time"],axis=1,inplace=True)
# """==========================================================================================================
# 4 time时间归一化 小时
# """
# data['time'] = pd.to_datetime(data['time'])
# time_now = data['time'].apply(lambda x:int((x-datetime(2018,11,14,0,0,0)).seconds/3600))
# data['time']= time_now
"""==========================================================================================================
2 划分训练集和验证集,验证集比例为test_size
"""
print("划分训练集和验证集,验证集比例为test_size")
train, test = train_test_split(data, test_size=0.1, random_state=666)
"""==========================================================================================================
3 提取lable 和特征
"""
y_train= train.status
train.drop(["status"],axis=1,inplace=True)
y_test= test.status
test.drop(["status"],axis=1,inplace=True)
"""==========================================================================================================
5 模型训练
"""
print("模型训练")
lg = LogisticRegression(C=100,dual=True)
lg.fit(train,y_train)
"""
保存模型
"""
print('保存模型')
joblib.dump(lg, path + "lg_120.m")
"""==========================================================================================================
6 模型预测
"""
y_test_pre = lg.predict(test)
"""==========================================================================================================
7 模型评分
"""
score_vali = f1_score(y_test, y_test_pre, average='macro')
print("验证集分数:{}".format(score_vali))
评分结果
需要改进的地方
- 交叉验证
- 网格搜索调参
- 模型融合
- 特征工程