CODEWITHSUNDEEP
python-3.xpandasgroup-by
misguided
misguided

Reputation: 3789

dataframe group by month count

I have created a dataframe as below:

paystart = datetime.date(2017, 10, 26)
paydate = pd.DataFrame()
paydate['PayDate'] = pd.date_range(paystart, end, freq='14D')
print(paydate.Grouper(freq='M'))

I want to count the number instances of date for any month-year combination i.e. Result should look like:

 2017-10   1
 2017-11   2
 2017-12   2

Upvotes: 1

Views: 47

Answers (1)

jezrael
jezrael

Reputation: 862581

If use Grouper with GroupBy.size or DataFrame.resample with Resampler.size output is DatetimeIndex:

paydate = pd.DataFrame()
paydate['PayDate'] = pd.date_range('2017-10-26', '2017-12-26', freq='14D')
print (paydate)
     PayDate
0 2017-10-26
1 2017-11-09
2 2017-11-23
3 2017-12-07
4 2017-12-21

print(paydate.groupby(pd.Grouper(freq='M', key='PayDate')).size())
PayDate
2017-10-31    1
2017-11-30    2
2017-12-31    2
Freq: M, dtype: int64

print(paydate.resample('M', on='PayDate').size())
PayDate
2017-10-31    1
2017-11-30    2
2017-12-31    2
Freq: M, dtype: int64

Or is possible create month periods by Series.dt.to_period - output is PeriodIndex:

print(paydate.groupby(paydate['PayDate'].dt.to_period('m')).size())
PayDate
2017-10    1
2017-11    2
2017-12    2
Freq: M, dtype: int64

Upvotes: 1

Related Questions