Hive: error calculating SUM then MAX of grouped items - sql

Question

I would like to run a query that calculates maximum money spent for each month of each credit card. For each credit card, I will need to calculate the sum of money spent each month. I have a table containing transactions of credit cards credit_transact:
processdate timestamp ""
cardno_hash string ""
amount int ""
year int ""
month int ""
Made-up sample data:
card year month amount
a123 2016 12 23160
a123 2016 10 287
c123 2016 11 5503
c123 2016 11 4206
I would like:
card year month amount
a123 2016 12 23160
c123 2016 11 9709
One important thing is year and month are partition columns.
I have tried a subquery like below:
USE credit_card_db;
SELECT sum_amount_transact.cardno_hash, sum_amount_transact.year, sum_amount_transact.month, MAX(sum_amount_transact.sum_amount)
FROM
(
SELECT cardno_hash, year, month, SUM(amount) AS sum_amount FROM credit_transact
GROUP BY cardno_hash, year, month
) AS sum_amount_transact
GROUP BY sum_amount_transact.cardno_hash, sum_amount_transact.year;
However, the following error is shown:
java.lang.Exception: org.apache.hive.service.cli.HiveSQLException: Error while compiling statement: FAILED: SemanticException Line 0:-1 Invalid column reference 'month'
The following subquery worked fine and returned results as expected:
SELECT cardno_hash, year, month, SUM(amount) AS sum_amount FROM credit_transact
GROUP BY cardno_hash, year, month
The result is:
card year month amount
a123 2016 12 23160
a123 2016 10 287
c123 2016 11 9709
Would very much appreciate if anyone can help with this problem.

Answer 1

I can't quite tell what you really want, but I'm pretty sure you want row_number(). I think you want the maximum month per year:
SELECT ct.*
FROM (SELECT cardno_hash, year, month, SUM(amount) AS sum_amount,
ROW_NUMBER() OVER (PARTITION BY cardno_hash, year ORDER BY SUM(amount) DESC) as seqnum
FROM credit_transact
GROUP BY cardno_hash, year, month
) ct
WHERE seqnum = 1;