按组和插值扩展缺少日期的数据框

Question

我有一个这样的数据框：

df = pd.DataFrame({"ID":["A", "A", "A" ,"B", "B"], "date":["06/24/2014","06/26/2014","06/29/2014","07/02/1999","07/04/1999"], "value": ["4","6","9","2","4"] })


ID        date value
0  A  06/24/2014     4
1  A  06/26/2014     6
2  A  06/29/2014     9
3  B  07/02/1999     2
4  B  07/04/1999     4

对于每个组，我想扩展数据框以包括日期的最大值和最小值之间的所有缺失日期，然后线性插值列值。结果应如下所示：

 ID        date value
0  A  06/24/2014     4
1  A  06/25/2014     5
2  A  06/26/2014     6
3  A  06/27/2014     7
4  A  06/28/2014     8
5  A  06/29/2014     9
6  B  07/02/1999     2
7  B  07/03/1999     3
8  B  07/04/1999     4

到目前为止我的想法如下：

将日期设为索引：

df.date = pd.DatetimeIndex(df.date)

按ID分组并应用以下功能：

B = df1.groupby('ID').apply(lambda x: x.reindex(pd.date_range(x.date.min(),x.date.max()), fill_value=0) )

最好的办法是什么？

谢谢，

Answer 1

我必须做一些初步调节以确保正确的dtypes

设置

df = pd.DataFrame({"ID":["A", "A", "A" ,"B", "B"],
                   "date":["06/24/2014","06/26/2014","06/29/2014","07/02/1999","07/04/1999"],
                   "value": ["4","6","9","2","4"] })


df.date = pd.to_datetime(df.date)
df.value = pd.to_numeric(df.value, 'coerce')

df = df.set_index('date')

解决方案

df.groupby('ID', group_keys=False).value \
    .apply(lambda df: df.resample('D').interpolate()).reset_index()

Answer 2

我这样做：

In [6]: df.groupby('ID').apply(lambda x: x.set_index('date').resample('D').pad())
Out[6]:
              ID value
ID date
A  2014-06-24  A     4
   2014-06-25  A     4
   2014-06-26  A     6
   2014-06-27  A     6
   2014-06-28  A     6
   2014-06-29  A     9
B  1999-07-02  B     2
   1999-07-03  B     2
   1999-07-04  B     4