更改DataFrame中的索引数量？

Question

我试图更改以下代码的输出:

import numpy as np
import pandas as pd
from pandas import Series, DataFrame, Panel, bdate_range, DatetimeIndex, date_range
from pandas.tseries.holiday import get_calendar
from datetime import datetime, timedelta
import pytz as pytz
from pytz import timezone

start =  datetime(2013, 1, 1)

hr1 = np.loadtxt("Spot_2013_Hour1.txt")

index = date_range(start, end = '2013-12-31', freq='B')
Allhrs = Series(index)
Allhrs = DataFrame({'hr1': hr1})
df = Allhrs
indexed_df = df.set_index(index)
print indexed_df

错误:

  File "<ipython-input-61-c7890d8ccb07>", line 17, in <module>
    indexed_df = df.set_index(index)

  File "/Applications/anaconda/lib/python2.7/site-packages/pandas/core/frame.py", line 2390, in set_index
    frame.index = index

  File "/Applications/anaconda/lib/python2.7/site-packages/pandas/core/generic.py", line 1849, in __setattr__
    object.__setattr__(self, name, value)

  File "properties.pyx", line 65, in pandas.lib.AxisProperty.__set__ (pandas/lib.c:38491)

  File "/Applications/anaconda/lib/python2.7/site-packages/pandas/core/generic.py", line 400, in _set_axis
    self._data.set_axis(axis, labels)

  File "/Applications/anaconda/lib/python2.7/site-packages/pandas/core/internals.py", line 1965, in set_axis
    'new values have %d elements' % (old_len, new_len))

ValueError: Length mismatch: Expected axis has 365 elements, new values have 261 elements

问题:

我有一个时间序列,我从一个txt文件加载.时间序列由365个元素组成,即2013年的所有日子.我需要这个txt文件,因为我需要分析每一天.

此外,我需要分析2013年的特定日期.所以我想改变数据的读数,即我想只看到工作日.此外,看到/打印特定的日子会很棒.

帮助赞赏

Answer 1

首先,创建一年中所有日期的DataFrame(或系列):

index = date_range(start='2013-1-1', end='2013-12-31', freq='D')
df = pd.DataFrame(hr1, index=index)

接下来,使用df.asfreq('B')下采样df到工作日:

import numpy as np
import pandas as pd

# hr1 = np.loadtxt("Spot_2013_Hour1.txt")
hr1 = np.random.random(365)
index = date_range(start='2013-1-1', end='2013-12-31', freq='D')
df = pd.DataFrame(hr1, index=index)

indexed_df = df.asfreq('B')
print(indexed_df)

要将频率设置为工作日,同时排除特定日期,您可以使用offsets.CustomBusinessDay:

import pandas.tseries.offsets as offsets
holidays = ['2013-10-03' , '2013-12-25']
business_days = offsets.CustomBusinessDay(holidays=holidays)
custom_df = df.asfreq(business_days)

因此,custom_df比两天少了indexed_df

In [12]: len(custom_df)
Out[12]: 259

In [13]: len(indexed_df)
Out[13]: 261

并且'2013-10-03'缺少"假期" :

In [18]: '2013-10-03' in indexed_df.index
Out[18]: True

In [19]: '2013-10-03' in custom_df.index
Out[19]: False

知道该reindex方法可用于子选择行也很有用.例如,您可以从indexed_df.index以下位置减去特定日期:

idx = indexed_df.index - pd.DatetimeIndex(holidays)
custom_df2 = df.reindex(idx)

结果,custom_df2等于custom_df:

In [35]: custom_df2.equals(custom_df)
Out[35]: True

但请注意索引有点不同:

In [36]: custom_df.index
Out[36]: 
<class 'pandas.tseries.index.DatetimeIndex'>
[2013-01-01, ..., 2013-12-31]
Length: 259, Freq: C, Timezone: None

In [37]: custom_df2.index
Out[37]: 
<class 'pandas.tseries.index.DatetimeIndex'>
[2013-01-01, ..., 2013-12-31]
Length: 259, Freq: None, Timezone: None

该custom_df作为Freq: C,而custom_df2有Freq: None.的freq是由某些方法中,如用于snap和to_period.但是这些方法也允许你指定所需的频率作为参数,所以在实践中我没有发现这个差异是一个大问题.