Aggregation based on another column values - sql

Question

Assume I have the following table:
+------------+-------------+
| Product_id | customer_id |
+------------+-------------+
| a | c1 |
| a | c2 |
| a | c3 |
| a | c4 |
| b | c1 |
| c | c1 |
| b | c2 |
| d | c2 |
+------------+-------------+
I want to find the number of (a, b, c) products purchases per customer and the number of (a, b, d) products purchases per customer. I tried to use COUNT with GROUP BY but I only managed to the find the number purchases of each customers FIDDLE. Do I need to use CASE WHEN or DECODE? How can I achieve that?
The expected output is something like:
+-------------+-------------+-------------+
| CUSTOMER_ID | ABC_PRODUCT | ABD_PRODUCT |
+-------------+-------------+-------------+
| c1 | 1 | 0 |
| c2 | 0 | 1 |
| c3 | 0 | 0 |
| c4 | 0 | 0 |
+-------------+-------------+-------------+

Answer 1

You can do this with a single aggregation and no subqueries. The key is using a nested case statement with aggregation to count each product for each customer. The following determines whether a customer has each "bundle":
SELECT CUSTOMER_ID,
(case when max(case when product_id = 'a' then 1 else 0 end) +
max(case when product_id = 'b' then 1 else 0 end) +
max(case when product_id = 'c' then 1 else 0 end) = 3
then 1
else 0
end) as ABC,
(case when max(case when product_id = 'a' then 1 else 0 end) +
max(case when product_id = 'b' then 1 else 0 end) +
max(case when product_id = 'd' then 1 else 0 end) = 3
then 1
else 0
end) as ABD
FROM CUSTOMERS_SALES
GROUP BY CUSTOMER_ID;
Now, your question is actually about the number of such purchases. So, I suppose a customer could purchase each item twice, and you would want them counted twice. If so, then the number is the least value of any counts. You can get this as well:
SELECT CUSTOMER_ID,
least(sum(case when product_id = 'a' then 1 else 0 end),
sum(case when product_id = 'b' then 1 else 0 end),
sum(case when product_id = 'c' then 1 else 0 end)
) as ABC,
least(sum(case when product_id = 'a' then 1 else 0 end),
sum(case when product_id = 'b' then 1 else 0 end),
sum(case when product_id = 'd' then 1 else 0 end)
) as ABD
FROM CUSTOMERS_SALES
GROUP BY CUSTOMER_ID;

Answer 2

Please try below query to find customer having the products a, b and c:
SELECT CUSTOMER_ID
FROM CUSTOMERS_SALES
WHERE PRODUCT_ID IN ('a', 'b', 'c')
GROUP BY CUSTOMER_ID
HAVING COUNT(DISTINCT PRODUCT_ID)=3
To get the count try:
SELECT COUNT(*) FROM(
SELECT CUSTOMER_ID
FROM CUSTOMERS_SALES
WHERE PRODUCT_ID IN ('a', 'b', 'd')
GROUP BY CUSTOMER_ID
HAVING COUNT(DISTINCT PRODUCT_ID)=3
)x

Answer 3

Based entirely off #TechDo's example:
SELECT DISTINCT CUSTOMER_ID,
DECODE((SELECT CUSTOMER_ID
FROM CUSTOMERS_SALES CS2
WHERE PRODUCT_ID IN ('A', 'B', 'C')
AND CS2.CUSTOMER_ID = CS.CUSTOMER_ID
GROUP BY CUSTOMER_ID
HAVING COUNT(DISTINCT PRODUCT_ID)=3), NULL, 0, 1) AS ABC_PRODUCT,
DECODE((SELECT CUSTOMER_ID
FROM CUSTOMERS_SALES CS2
WHERE PRODUCT_ID IN ('A', 'B', 'D')
AND CS2.CUSTOMER_ID = CS.CUSTOMER_ID
GROUP BY CUSTOMER_ID
HAVING COUNT(DISTINCT PRODUCT_ID)=3), NULL, 0, 1) AS ABD_PRODUCT
FROM CUSTOMERS_SALES CS
ORDER BY CUSTOMER_ID

Answer 4

SELECT CUSTOMERS_SALES.CUSTOMER_ID,NVL(MAX(abc.CUSTOMER_ID),0) as ABC_PRODUCT ,NVL(MAX(abd.CUSTOMER_ID),0) as ABD_PRODUCT FROM CUSTOMERS_SALES
LEFT JOIN
(SELECT CUSTOMER_ID
FROM CUSTOMERS_SALES
WHERE PRODUCT_ID IN ('a', 'b', 'd')
GROUP BY CUSTOMER_ID
HAVING COUNT(DISTINCT PRODUCT_ID)=3) abd
ON abd.CUSTOMER_ID=CUSTOMERS_SALES.CUSTOMER_ID
LEFT JOIN
(SELECT CUSTOMER_ID
FROM CUSTOMERS_SALES
WHERE PRODUCT_ID IN ('a', 'b', 'c')
GROUP BY CUSTOMER_ID
HAVING COUNT(DISTINCT PRODUCT_ID)=3) abc
ON abc.CUSTOMER_ID=CUSTOMERS_SALES.CUSTOMER_ID
GROUP BY CUSTOMERS_SALES.CUSTOMER_ID
ORDER BY CUSTOMERS_SALES.CUSTOMER_ID;
FIDDLE