如何使用groupby / cut将Pandas DataFrame日期分组到自定义日期范围区间

Question

我正在尝试使用groupby和cut将自定义范围的日期分组到目前为止没有成功。从返回的错误消息中，我想知道cut是否正在尝试将我的日期作为数字处理。

我想按自定义日期范围对df1['date']进行分组，然后对df1['HDD']值求和。自定义范围可在df2中找到：

import pandas as pd
df1 = pd.DataFrame( {'date': ['2/1/2015', '3/2/2015', '3/3/2015', '3/4/2015','4/17/2015','5/12/2015'],
                             'HDD' : ['7.5','8','5','23','11','55']})
    HDD  date
0   7.5 2/1/2015
1   8   3/2/2015
2   5   3/3/2015
3   23  3/4/2015
4   11  4/17/2015
5   55  5/12/2015

df2有自定义日期范围：

df2 = pd.DataFrame( {'Period': ['One','Two','Three','Four'],
                     'Start Dates': ['1/1/2015','2/15/2015','3/14/2015','4/14/2015'],
                     'End Dates' : ['2/14/2015','3/13/2015','4/13/2015','5/10/2015']})

    Period  Start Dates End Dates
0   One     1/1/2015    2/14/2015
1   Two     2/15/2015   3/13/2015
2   Three   3/14/2015   4/13/2015
3   Four    4/14/2015   5/10/2015

我想要的输出是按照自定义日期范围对df1进行分组，并聚合每个Period的HDD值。应该输出这样的东西：

   Period    HDD
0  One       7.5
1  Two       36
2  Three     0
3  Four      11

以下是我尝试使用自定义分组的一个示例：

df3 = df1.groupby(pd.cut(df1['date'], df2['Start Dates'])).agg({'HDD': sum})

......这是我得到的错误：

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-103-55ea779bcd73> in <module>()
----> 1 df3 = df1.groupby(pd.cut(df1['date'], df2['Start Dates'])).agg({'HDD': sum})

/opt/conda/lib/python3.5/site-packages/pandas/tools/tile.py in cut(x, bins, right, labels, retbins, precision, include_lowest)
    112     else:
    113         bins = np.asarray(bins)
--> 114         if (np.diff(bins) < 0).any():
    115             raise ValueError('bins must increase monotonically.')
    116 

/opt/conda/lib/python3.5/site-packages/numpy/lib/function_base.py in diff(a, n, axis)
   1576         return diff(a[slice1]-a[slice2], n-1, axis=axis)
   1577     else:
-> 1578         return a[slice1]-a[slice2]
   1579 
   1580 

TypeError: unsupported operand type(s) for -: 'str' and 'str'

• 是否试图将我的日期范围作为数字处理？
• 我是否需要明确地将我的日期转换为datetime对象（试过这个但是可能是正确的）？

感谢您提出的任何建议！

Answer 1

如果您将所有日期从dtype字符串转换为datetime，则此方法有效。

df1['date'] = pd.to_datetime(df1['date'])

df2['End Dates'] = pd.to_datetime(df2['End Dates'])

df2['Start Dates'] = pd.to_datetime(df2['Start Dates'])

df1['HDD'] = df1['HDD'].astype(float)

df1.groupby(pd.cut(df1['date'], df2['Start Dates'])).agg({'HDD': sum})

输出：

                           HDD
date                          
(2015-01-01, 2015-02-15]   7.5
(2015-02-15, 2015-03-14]  36.0
(2015-03-14, 2015-04-14]   NaN

添加标签：

df1.groupby(pd.cut(df1['date'], df2['Start Dates'], labels=df2.iloc[:-1,1])).agg({'HDD': sum})

输出：

        HDD
date       
One     7.5
Two    36.0
Three   NaN