Pandas DataFrame按天/小时/分钟切片

Question

我有一个带有日期时间索引的pandas Dataframe,如'YYYY-MM-DD HH:MM:SS'.

Index               Parameter
2007-05-02 14:14:08     134.8
2007-05-02 14:14:32     134.8 
2007-05-02 14:14:41     134.8 
2007-05-02 14:14:53     134.8 
2007-05-02 14:15:01     134.8 
2007-05-02 14:15:09     134.8 
......
2007-05-30 23:08:02     105.9 
2007-05-30 23:18:02     105.9 
2007-05-30 23:28:02     105.9 
2007-05-30 23:38:03     105.8 

可以按年df['2007']或按月分割数据帧df['2007-05']吗？

但是,当我试图在白天切片DataFrame时df['2007-05-02'],我得到了错误:

KeyError: < Timestamp: 2007-02-05 00:00:00. 

我使用pandas版本8.0.1.是否可以以比年或月更小的频率切片DataFrame？例如,按天或小时？

Answer 1

使用df.ix[x:y]where x和y是datetime对象.

例:

In [117]: frame.index.summary()
Out[117]: 'DatetimeIndex: 6312960 entries, 2000-04-05 00:01:00 to 2012-04-06 00:00:00\nFreq: T'


In [118]: x=datetime(2001, 4, 5, 0, 1)

In [119]: y=datetime(2001, 4, 5, 0, 5)

In [120]: print frame.ix[x:y]
                     radiation      tamb
2001-04-05 00:01:00  67.958873  8.077386
2001-04-05 00:02:00  50.801294  0.731453
2001-04-05 00:03:00  16.042035  6.944998
2001-04-05 00:04:00   5.678343  9.728967
2001-04-05 00:05:00  72.551601  7.652942

你也可以这样做:

In [121]: print frame.ix[x]
radiation    67.958873
tamb          8.077386
Name: 2001-04-05 00:01:00