在 Pandas 中使用替换同时填充和填充

Question

Question

Is there a way to ffill and bfill at the same time using replace in Pandas?

See the following example :

l = 12
rng = pd.date_range('1/1/2011', periods=l, freq='8h')
df = pd.DataFrame({ 
    'animals':[0,0,'cat',0,'dog',0,0,0,'mouse',0,'ant',0],
    },index=rng)

df
Out[93]:
                  animals
2011-01-01 00:00:00 0
2011-01-01 08:00:00 0
2011-01-01 16:00:00 cat
2011-01-02 00:00:00 0
2011-01-02 08:00:00 dog
2011-01-02 16:00:00 0
2011-01-03 00:00:00 0
2011-01-03 08:00:00 0
2011-01-03 16:00:00 mouse
2011-01-04 00:00:00 0
2011-01-04 08:00:00 ant
2011-01-04 16:00:00 0

Im currently doing this using two iterations of replace .

df.animals = df.groupby(df.index.to_datetime().day).transform(lambda x: x.replace(to_replace=0, method='ffill'))
df.animals = df.groupby(df.index.to_datetime().day).transform(lambda x: x.replace(to_replace=0, method='bfill'))

df

                  animals
2011-01-01 00:00:00 cat
2011-01-01 08:00:00 cat
2011-01-01 16:00:00 cat
2011-01-02 00:00:00 dog
2011-01-02 08:00:00 dog
2011-01-02 16:00:00 dog
2011-01-03 00:00:00 mouse
2011-01-03 08:00:00 mouse
2011-01-03 16:00:00 mouse
2011-01-04 00:00:00 ant
2011-01-04 08:00:00 ant
2011-01-04 16:00:00 ant
    ​

It works ok but I thought there may be a way to ffill and bfill so thought worth checking on SO.

Answer 1

IIUC you can do it this way:

In [278]: df['animals'] = df['animals'].replace(0, np.nan) \
                                       .groupby(pd.TimeGrouper('D')) \
                                       .bfill().ffill()

In [279]: df
Out[279]:
                    animals
2011-01-01 00:00:00     cat
2011-01-01 08:00:00     cat
2011-01-01 16:00:00     cat
2011-01-02 00:00:00     dog
2011-01-02 08:00:00     dog
2011-01-02 16:00:00     dog
2011-01-03 00:00:00   mouse
2011-01-03 08:00:00   mouse
2011-01-03 16:00:00   mouse
2011-01-04 00:00:00     ant
2011-01-04 08:00:00     ant
2011-01-04 16:00:00     ant