Going forward, interpolate
works great:
name days
0 a NaN
1 a NaN
2 a 2
3 a 3
4 a NaN
5 a NaN records.loc[:, 'days'].interpolate(method='linear', inplace=True)name days
0 a NaN
1 a NaN
2 a 2
3 a 3
4 a 4
5 a 5
...however, it does not address the beginning rows (only goes forward). The limit_direction
param allows {‘forward’, ‘backward’, ‘both’}
. None of these works. Is there a proper way to interpolate backwards?
We can assume a series incrementing or decrementing by 1, which may not start at 0 as it happens to in this example.
It seems it works only with parameter limit
see docs [In 47]:
Add a limit_direction keyword argument that works with limit to enable interpolate to fill NaN values forward, backward, or both (GH9218, GH10420, GH11115)
records = pd.DataFrame(
{'name': {0: 'a', 1: 'a', 2: 'a', 3: 'a', 4: 'a', 5: 'a', 6: 'a', 7: 'a', 8: 'a', 9: 'a'},
'days': {0: 0.0, 1: np.nan, 2: np.nan, 3: np.nan, 4: 4.0, 5: 5.0, 6: np.nan, 7: np.nan, 8: np.nan, 9: 9.0}},
columns=['name','days'])print (records)name days
0 a 0.0
1 a NaN
2 a NaN
3 a NaN
4 a 4.0
5 a 5.0
6 a NaN
7 a NaN
8 a NaN
9 a 9.0
#by default limit_direction='forward'
records['forw'] = records['days'].interpolate(method='linear', limit=1)
records['backw'] = records['days'].interpolate(method='linear',limit_direction='backward', limit=1)
records['both'] = records['days'].interpolate(method='linear', limit_direction='both', limit=1)
print (records)name days forw backw both
0 a 0.0 0.0 0.0 0.0
1 a NaN 1.0 NaN 1.0
2 a NaN NaN NaN NaN
3 a NaN NaN 3.0 3.0
4 a 4.0 4.0 4.0 4.0
5 a 5.0 5.0 5.0 5.0
6 a NaN 6.0 NaN 6.0
7 a NaN NaN NaN NaN
8 a NaN NaN 8.0 8.0
9 a 9.0 9.0 9.0 9.0