pandas groupby agg get max from one column and bring along value from another column-CodePudding

I have a dataframe of purchases:

product_id    count    timestamp           customer_id
1             1        2021-10-04 10:20    a
1             3        2021-10-04 10:21    b
2             4        2021-10-04 10:00    c
1             2        2021-10-03 10:00    c

I use the following groupby and agg to create a report of the sum and mean of the count, and the latest purchase timestamp.

report = (
    df.groupby(product_id).agg(
        sum=pd.NamedAgg(column="count", aggfunc="sum"),
        mean_count=pd.NamedAgg(column="count", aggfunc="mean"),
        latest_purchase_time=pd.NamedAgg(column="timestamp", aggfunc="max")
    )
)

I want to include in this report the customer_id that corresponds to the latest purchase timestamp. Is there a way to do this?

e.g.:

product_id    sum    mean_count    latest_purchase_timestamp   *customer_id*
1             6      2             2021-10-04 10:21            b
2             4      4             2021-10-04 10:00            c

CodePudding user response：

First convert customer_id to index, so possible get value by maximal timestamp by idxmax:

report = (
    df.set_index('customer_id')
      .groupby('product_id').agg(
        sum=pd.NamedAgg(column="count", aggfunc="sum"),
        mean_count=pd.NamedAgg(column="count", aggfunc="mean"),
        latest_purchase_time=pd.NamedAgg(column="timestamp", aggfunc="max"),
        customer_id=pd.NamedAgg(column="timestamp", aggfunc="idxmax")
    )
)
print (report)
            sum  mean_count latest_purchase_time customer_id
product_id                                                  
1             6           2  2021-10-04 10:21:00           b
2             4           4  2021-10-04 10:00:00           c