group by - working with groupby multiple columns in Python -

i have 3 columns in dataframe, as:

user_id, product_category_1 , corresponding purchase amount.

i trying group based on user_id , product_category_1 , selecting average of purchase amount.

so output dataframe have: user_id,product_category_1 , avg_purchase.

this not working me:

x=train_bk.groupby(["user_id","product_category_1"],as_index=false)['purchase'].transform('mean') 

this gives me series of mean value of purchase each row. need keep unique user_id , product_category_1 combination

   x1 = train_bk.select(average(train_bk.user_id), train_bk.product_category_1,           group_by=(train_bk.user_id,train_bk.product_category_1)) 

i tried sql package. throws error: "name 'average' not defined". there package in python has sql syntax similar teradata or mysql.

ok, seems working:

   x = train_bk.groupby(["user_id","product_category_1"],as_index=false)['purchase'].mean()