如何使用过去几个月的最新可用数据来填充缺失值？

Question

I have a dataframe like this:

Month/Year  Value   
01/2018     100 
03/2018     200 
06/2018     500 

The values for 02/2018, 04/2018 and 05/2018 is missing because the value did not change in those months. I would like to have a dataframe which incudes the missing months:

Month/Year  Value   
01/2018     100 
02/2018     100 
03/2018     200 
04/2018     200 
04/2018     200 
06/2018     500 

Can anyone help?

Answer 1

One way to do this:

df.assign(**{"Month/Year": pd.to_datetime(data["Month/Year"])}).set_index("Month/Year").resample("M").ffill().reset_index()

Should yield:

  Month/Year  Value
0 2018-01-31    100
1 2018-02-28    100
2 2018-03-31    200
3 2018-04-30    200
4 2018-05-31    200
5 2018-06-30    500

df here is your starting dataframe. It gets resampled to a monthly frequency and we use the .ffill method to fill the values for the missing months.

I opted for a one-liner but you can break it down to a more structured block of code. You can also reformat the Month/Year column after the resampling.

I hope this helps.

Answer 2

You can use pd.DataFrame.resample , then pd.Series.ffill to forward-fill null values. If you require string dates, an extra conversion is required, as below.

df['date'] = pd.to_datetime(df['Month/Year'])

res = df.resample('M', on='date')\
        .sum().ffill().astype(int)\
        .reset_index()

res['date'] = res['date'].dt.strftime('%m/%Y')

print(res)

      date  Value
0  01/2018    100
1  02/2018    100
2  03/2018    200
3  04/2018    200
4  05/2018    200
5  06/2018    500