python数据挖掘（4.决策树）

数据源

本章介绍另一种分类算法：决策树，比起其他算法决策树最主要的一个优点诗决策过程是机器和人都能看懂的，我们使用机器学习到的模型就能完成预测任务，另一个优点是他可以处理多种不同类型的特征。
我们这章使用的数据请在文章开头的数据源中python数据挖掘/Chapter4中的文件

这一章的数据诗NBA2013-2014赛季的比赛数据，这是一个CSV文件，我们将它读取到pandas中看一下

In [1]: import numpy as np                                                      

In [2]: import pandas as pd                                                     

In [3]: dataset = pd.read_csv('leagues_NBA_2014_gam
   ...: es_games.csv')                                                          

In [4]: dataset.head()                                                          
Out[4]: 
              Date        NaN       Visitor/Neutral  ...  PTS.1 NaN.1  Notes
0  Tue Oct 29 2013  Box Score         Orlando Magic  ...     97   NaN    NaN
1  Tue Oct 29 2013  Box Score  Los Angeles Clippers  ...    116   NaN    NaN
2  Tue Oct 29 2013  Box Score         Chicago Bulls  ...    107   NaN    NaN
3  Wed Oct 30 2013  Box Score         Brooklyn Nets  ...     98   NaN    NaN
4  Wed Oct 30 2013  Box Score         Atlanta Hawks  ...    118   NaN    NaN

现在这个数据有一些问题

1. 第一列Date日期是字符串
2. 表头需要优化

那么我们从新搞一哈
幸亏pandas可以将很多种字符串日期转化为标准日期对象

In [5]: dataset = pd.read_csv('/Users/gn/scikit--learn/data/leagues_NBA_2014_gam
   ...: es_games.csv',parse_dates = ['Date'],skiprows=[0,])               
         
In [6]: dataset.columns = ['Data','Scire Type','Visitor Team','VisitorPts','Home
   ...:  Team','HomePts','OT','Notes']   
   
In [7]: dataset.head()                                                                                                                                                                                     
Out[7]: 
        Data Scire Type          Visitor Team  VisitorPts            Home Team  HomePts   OT Notes
0 2013-10-29  Box Score         Orlando Magic          87       Indiana Pacers       97  NaN   NaN
1 2013-10-29  Box Score  Los Angeles Clippers         103   Los Angeles Lakers      116  NaN   NaN
2 2013-10-29  Box Score         Chicago Bulls          95           Miami Heat      107  NaN   NaN
3 2013-10-30  Box Score         Brooklyn Nets          94  Cleveland Cavaliers       98  NaN   NaN
4 2013-10-30  Box Score         Atlanta Hawks         109     Dallas Mavericks      118  NaN   N

现在看起来是不是好多了
由于数据中不包含胜负数据，需要我们将比分转化为直观的胜负
现在需要创建一些特征用于数据挖掘，我们使用上一场在主场的胜负和在客场的胜负来判断

In [11]: dataset['HomeWin'] = dataset['VisitorPts'] < dataset['HomePts']                                                                            

In [12]: from collections import defaultdict                                                                                                        

In [13]: won_last = defaultdict(int)                                                                                                                

In [14]: dataset["HomeLastWin"] = False                                                                                                             

In [15]: dataset["VisitorLastWin"] = False                                                                                                          

In [16]: for index, row in dataset.iterrows(): 
    ...:     home_team = row["Home Team"] 
    ...:     visitor_team = row["Visitor Team"] 
    ...:     row["HomeLastWin"] = won_last[home_team] 
    ...:     row["VisitorLastWin"] = won_last[visitor_team] 
    ...:     dataset.loc[index] = row  
    ...:     won_last[home_team] = row["HomeWin"] 
    ...:     won_last[visitor_team] = not row["HomeWin"] 
    ...:                                                                                                                                            

In [17]: dataset.loc[20:25]                                                                                                                         
Out[17]: 
         Data Scire Type            Visitor Team  VisitorPts           Home Team  HomePts   OT Notes  HomeWin HomeLastWin VisitorLastWin
20 2013-11-01  Box Score         Milwaukee Bucks         105      Boston Celtics       98  NaN   NaN    False       False          False
21 2013-11-01  Box Score              Miami