How can I increment a date in datetime64[ns] format by 6 calendar months?

Question

I never realised that incrementing a simple date in Python would be such an insurmountable challenge, but have given up after 2 hours of trying and searching on this forum. I have a dataframe with a column effective_date, which contains entries like 2019-01-02 and datatype datetime64[ns].

I've tried:

data['effective_date'] = pd.to_datetime(data['effective_date'].values)
data['six_mth_interval'] = data['effective_date'].apply(lambda x: x['effective_date'].values + relativedelta(months=6))

... but I get the following error:

<ipython-input-315-b81c59eb6b0d> in <lambda>(x)
----> 1 data['six_mth_interval'] = data['effective_date'].apply(lambda x: x['effective_date'] + relativedelta(months=6))

TypeError: 'Timestamp' object is not subscriptable

Existing articles on S/O have not been helpful.

Answer 1

Run:

data['six_mth_interval'] = data.effective_date + pd.DateOffset(months=6)

For an example DataFrame the result is:

   effective_date six_mth_interval
0      2019-08-01       2020-02-01
1      2019-08-02       2020-02-02
2      2019-08-25       2020-02-25
3      2019-08-26       2020-02-26
4      2019-08-27       2020-02-27
5      2019-08-28       2020-02-28
6      2019-08-29       2020-02-29
7      2019-08-30       2020-02-29
8      2019-08-31       2020-02-29
9      2019-09-01       2020-03-01
10     2019-09-02       2020-03-02

Your code failed because when you apply a function to a Series (a column of a DataFrame), then the argument is each single element of this Series.

So you could write:

data['six_mth_interval'] = data['effective_date'].apply(
    lambda x: x + pd.DateOffset(months=6))

but my code is faster.