Question

So I have these specific columns that I’m working with:
customer_token, merchant_id, merchant_category_code, and transaction_amount.

My current query is this:

SELECT customer_token, COUNT(transaction_amount), SUM(transaction_amount)
FROM transaction 
                     WHERE file_date>20121031 
                     and file_date<20121201
GROUP BY customer_token

I want add on to the above query a part where in the result, the merchant_category_code is separated into different columns based on the transaction amount in each specific merchant_category_code. A result that would look something like this:

customer_token, count(transaction_amount),sum(transaction_amount), count(transaction_amount in merchant_category_code which is ranked 1), count(transaction_amount in merchant_category_code which is ranked 2), count(transaction_amount in merchant_category_code which is ranked 3), etc…

and then this:

customer_token, count(transaction_amount),sum(transaction_amount), sum(transaction_amount in merchant_category_code which is ranked 1), sum(transaction_amount in merchant_category_code which is ranked 2), sum(transaction_amount in merchant_category_code which is ranked 3), etc…

But I’m at a loss on how to do this or if it is even at all possible.

Answer 1

If you know in advance what the possible values of merchant_category_code are, you can use CASE expressions:

SELECT customer_token,
       COUNT(transaction_amount),
       SUM(transaction_amount),
       COUNT(CASE WHEN merchant_category_code = 1 THEN transaction_amount END),
       COUNT(CASE WHEN merchant_category_code = 2 THEN transaction_amount END),
       COUNT(CASE WHEN merchant_category_code = 3 THEN transaction_amount END),
       ...
       SUM(CASE WHEN merchant_category_code = 1 THEN transaction_amount END),
       SUM(CASE WHEN merchant_category_code = 2 THEN transaction_amount END),
       SUM(CASE WHEN merchant_category_code = 3 THEN transaction_amount END),
       ...
  FROM transaction 
 WHERE file_date BETWEEN 20121101 AND 20121130
 GROUP
    BY customer_token
;

(or IF expressions, if you prefer; for documentation on both of these, see the section titled “Conditional Functions” on the page “LanguageManual+UDF” in the Hive wiki).