CODEWITHSUNDEEP
pythonpandasdataframe
Dark Knight
Dark Knight

Reputation: 175

Pandas fill consecutive null date values from previous dates + a constant number of days

I have a dataframe that contains a data column 

Comp_date
0 2020-04-24
1        NaT
2        NaT
3        NaT
4 2020-08-06
5        NaT
6        NaT
7        NaT
8 2020-08-22
9        NaT

I am trying to fill the null with the value of the previous date + add a constant number of days (10). But I am unable to do so. I tried the following

df['Comp_date']=df['Comp_date'].fillna((df['Comp_date'].shift()+pd.to_timedelta(10, unit='D')), inplace=True)

Nothing happens and I get the same result. Any help? expected outcome 

    Comp_date
0   2020-04-24
1   2020-05-04
2   2020-05-14
3   2020-05-24
4   2020-08-06
5   2020-08-16
6   2020-08-26
7   2020-09-05
8   2020-08-22
9   2020-09-01

Upvotes: 1

Views: 857

Answers (2)

jezrael
jezrael

Reputation: 862511

Idea is create groups for missing values by Series.notna and Series.cumsum and create counter by GroupBy.cumcount, multiple number of days by Series.mul convert to timedeltas by to_timedelta what is added to forward filling missing values with ffill:

num_days = 10
g = df['Comp_date'].notna().cumsum()
days = pd.to_timedelta(df.groupby(g).cumcount().mul(num_days), unit='d')

df['Comp_date'] = df['Comp_date'].ffill().add(days)
print (df)
   Comp_date
0 2020-04-24
1 2020-05-04
2 2020-05-14
3 2020-05-24
4 2020-08-06
5 2020-08-16
6 2020-08-26
7 2020-09-05
8 2020-08-22
9 2020-09-01

Upvotes: 2

Alexander
Alexander

Reputation: 109528

I'm not clear on your question, but this adds a constant number of days to the last observed Comp_date.

constant_number_of_days = 2
df2 = df['Comp_date'].ffill().to_frame()
df2.loc[df['Comp_date'].isnull(), 'Comp_date'] += pd.Timedelta(days=constant_number_of_days)
>>> df2
   Comp_date
0 2020-04-24
1 2020-04-26
2 2020-04-26
3 2020-04-26
4 2020-08-06
5 2020-08-08
6 2020-08-08
7 2020-08-08
8 2020-08-22
9 2020-08-24

Upvotes: 2

Related Questions