如何使用Python / Pandas从“日期”字段按月分组

Question

I have a Dataframe df as follows:

date               value_1    value_2
2018.07.06           10          0
2018.07.14           20          1
2018.07.27           20          2
2018.08.06           30          1
2018.08.09           40          3
2018.08.13           20          2
2018.09.10           30          1
2018.09.22           50          2
2018.10.09           20          3
2018.10.27           20          1

I need to group the above data by month to get output as:

date              value_1    value_2
2018.07.01           50          3
2018.08.01           90          6
2018.09.01           80          3
2018.10.01           40          4

How can I do this efficiently in pandas?

Answer 1

Try, groupby using pd.Grouper with freq = 'MS':

df.groupby(pd.Grouper(freq='MS', key='date')).sum().reset_index()

Output:

        date  value_1  value_2
0 2018-07-01       50        3
1 2018-08-01       90        6
2 2018-09-01       80        3
3 2018-10-01       40        4

And, if you want get dot date format back, you can use this:

df_out = df.groupby(pd.Grouper(freq='MS', key='date')).sum().reset_index()

df_out['date'] = df_out['date'].dt.strftime('%Y.%m.%d')

df_out

Output:

         date  value_1  value_2
0  2018.07.01       50        3
1  2018.08.01       90        6
2  2018.09.01       80        3
3  2018.10.01       40        4

Answer 2

Do with

df.date=pd.to_datetime(df.date)
df.groupby(df.date+pd.offsets.MonthBegin(-1)).sum()
Out[171]: 
            value_1  value_2
date                        
2018-07-01       50        3
2018-08-01       90        6
2018-09-01       80        3
2018-10-01       40        4

Answer 3

If you have date as the index, it's as simple as resampling.

df.resample('MS').sum()

If you don't have it as the index alreay, you can set_index .

df.set_index('date').resample('MS').sum()

Both give you

            value_1  value_2
date                        
2018-07-01       50        3
2018-08-01       90        6
2018-09-01       80        3
2018-10-01       40        4

Answer 4

Use the dt accessor to get the months from the date column:

df = pd.read_csv(r'C:\Users\Tim\Desktop\data.txt')
df['date'] = pd.to_datetime(df['date'])
df.groupby(df['date'].dt.month).sum()

this will create the following output:

     value_1    value_2
date        
7   50  3
8   90  6
9   80  3
10  40  4