CODEWITHSUNDEEP
pythonpandasdataframe
Eli Turasky
Eli Turasky

Reputation: 1061

Pandas sum every value over certain month range every year

I want to sum data from January through June every year. I have a dataframe that looks like this:

Date       Value
1980-01-01 2
1980-02-01 3
1980-03-01 3
1980-04-01 2
1980-05-01 3
1980-06-01 3

I would like there to then be a new column that stores the value of the data as Sum for every 6 month interval each year such that the sum would = 16 for this example. I tried using a combination of df.groupby() and df.sum() but couldn't quite figure it out.

Expected output would look something like this:

Date       Value   Sum
1980-01-01 2       16
1980-02-01 3
1980-03-01 3
1980-04-01 2
1980-05-01 3
1980-06-01 3

Upvotes: 1

Views: 2137

Answers (1)

wwii
wwii

Reputation: 23753

idx = pd.Series(pd.date_range('1/1/2018', periods=100, freq='MS'),name='date')
df = pd.DataFrame(range(len(idx)), index=idx,columns=['A'])

Filter then resample.

>>> sums = df.loc[df.index.month.isin([1,2,3,4,5,6])].resample('YS').sum()
>>> sums['A'].values
array([ 15,  87, 159, 231, 303, 375, 447, 519, 390], dtype=int64)
>>> sums
              A
date           
2018-01-01   15
2019-01-01   87
2020-01-01  159
2021-01-01  231
2022-01-01  303
2023-01-01  375
2024-01-01  447
2025-01-01  519
2026-01-01  390
>>>

I assumed that date was the index in your example. If it is a column change you need to use the dt accessor in the filter and specify the column name in resample. 

dfa = pd.DataFrame({'date':idx,'A':range(len(idx))})
>>> sums = dfa.loc[dfa.date.dt.month.isin([1,2,3,4,5,6])].resample('YS',on='date').sum()
>>> sums['A'].values
array([ 15,  87, 159, 231, 303, 375, 447, 519, 390], dtype=int64)

You could also resample/aggregate on a six month frequency and just take every other result - it seems to work even if months are missing from the Series.

>>> dfq = dfa.loc[::2]
>>> dfq.head()
        date  A
0 2018-01-01  0
2 2018-03-01  2
4 2018-05-01  4
6 2018-07-01  6
8 2018-09-01  8
>>> dfc = dfq.resample('6MS', on='date').sum()
>>> dfc.loc[::2].head()
              A
date           
2018-01-01    6
2019-01-01   42
2020-01-01   78
2021-01-01  114
2022-01-01  150

If the DataFrame only contains the first six months of each year then you don't need to filter. Just resample.

>>> dfb = dfa.loc[dfa.date.dt.month.isin([1,2,3,4,5,6])]
>>> dfb.resample('YS',on='date').sum().head()
              A
date           
2018-01-01   15
2019-01-01   87
2020-01-01  159
2021-01-01  231
2022-01-01  303
>>>

Upvotes: 3

Related Questions