计算 Python pandas Dataframe 列并按日期排序

Question

I have a DataFrame that looks like;

                            ID        Tags       Priority 
Created At
2017-01-01 14:40:00        1234      some_tag     P1
2017-01-02 15:00:00        1345      more_tag     P4

I want to count all of the different priority tags for each week to look like;

Week       Priority     Count
1            p1           1
             p4           1

I've tried the fairly simplet things like:

print(df.groupby(df.index.date).count())

But that is not giving me what I need.

Any suggestions?

Answer 1

DateTimeIndex has a property called weekofyear which suits your needs. Furthermore, you do not only want to group on week, but also on priority. Luckily, the groupby function supports taking several attributes to group by:

import datetime
import pandas as pd

a = datetime.datetime.now()
b = datetime.datetime.now()
df = pd.DataFrame({'Created At': [a, b], 'ID': [1234, 1235], 'Priority': ['p1', 'p2']}).set_index('Created At')
print(df)

                              ID Priority
Created At                               
2017-02-08 22:25:51.862604  1234       p1
2017-02-08 22:25:57.095862  1235       p2

print(df.groupby([df.index.weekofyear, 'Priority']).count())

            ID
  Priority    
6 p1         1
  p2         1

Answer 2

Use pd.TimeGrouper with freq W-Sat so that 'Sunday' and 'Monday' get grouped together.

df.groupby([pd.TimeGrouper('W-Mon'), 'Priority']).ID.count().to_frame() 

                     ID
Created At Priority    
2017-01-07 P1         1
           P4         1