How to do a rolling count of values grouping by date python

user_17893180 0
Viewed 70

Hi I have a table of data like below and I want to try do a rolling count that takes the date in the group by and the values of dates prior.

Table of data:

Date ID
1/1/2020 123
2/1/2020 432
2/1/2020 5234
4/1/2020 543
5/1/2020 645
6/1/2020 231

My desired output is something like this:

Date count
1/1/2020 1
2/1/2020 3
4/1/2020 4
5/1/2020 5
6/1/2020 6

I have tried the following but it doesn't seem to work on how I want it do it.

df[['id','date']].groupby('date').cumcount()

pandas numpy python
1 Answers

Convert column to datetimes for correct ordering if aggregate GroupBy.size and add cumulative sum by Series.cumsum:

df['Date'] = pd.to_datetime(df['Date'], dayfirst=True)

df = df.groupby('Date').size().cumsum().reset_index(name='count')
print (df)
        Date  count
0 2020-01-01      1
1 2020-01-02      3
2 2020-01-04      4
3 2020-01-05      5
4 2020-01-06      6
user_29010020
Related