Reputation: 714
I have the following db table, and I would like to be able to count the instance of sales of certain products per salesperson.
|------------|------------|------------|
|id |user_id |product_id |
|------------|------------|------------|
|1 |1 |2 |
|2 |1 |4 |
|3 |1 |2 |
|4 |2 |1 |
|------------|------------|------------|
I would like to able to create a result set like the following;
|------------|-------------|------------|------------|------------|
|user_id |prod_1_count |prod_2_count|prod_3_count|prod_4_count|
|------------|-------------|------------|------------|------------|
|1 |0 |2 |0 |1 |
|2 |1 |0 |0 |0 |
|------------|-------------|------------|------------|------------|
I am creating graphs with this data, and once again (as earlier today) I am unable to count the column totals. I have tried;
SELECT user_id,
(SELECT count(product_id) FROM sales WHERE product_id = 1) AS prod_1_count,
(SELECT count(product_id) FROM sales WHERE product_id = 2) AS prod_2_count,
(SELECT count(product_id) FROM sales WHERE product_id = 3) AS prod_3_count,
(SELECT count(product_id) FROM sales WHERE product_id = 4) AS prod_4_count
FROM sales GROUP BY user_id;
I can see why this doesn't work, because for each bracketed SELECT the user_id doesn't match the external user_id in the main SELECT statement.
Upvotes: 51
Views: 72014
Reputation: 688
This is a very old question, but possibly this is another variant for other searchers.
Since MySQL 4.0
if()
can also be used at this point in combination with sum()
and ifnull()
. The ifnull()
function ensures that if there are no rows, a 0
is output per column.
select user_id,
ifnull(sum(if(product_id = 1, 1, 0)), 0) as prod_1_count,
ifnull(sum(if(product_id = 2, 1, 0)), 0) as prod_2_count,
ifnull(sum(if(product_id = 3, 1, 0)), 0) as prod_3_count,
ifnull(sum(if(product_id = 4, 1, 0)), 0) as prod_4_count
from your_table
group by user_id
Upvotes: 0
Reputation: 247720
You are trying to pivot the data. MySQL does not have a pivot function so you will have to use an aggregate function with a CASE
expression:
select user_id,
count(case when product_id = 1 then product_id end) as prod_1_count,
count(case when product_id = 2 then product_id end) as prod_2_count,
count(case when product_id = 3 then product_id end) as prod_3_count,
count(case when product_id = 4 then product_id end) as prod_4_count
from sales
group by user_id;
Upvotes: 31
Reputation: 65547
You can do this using SUM
and CASE
:
select user_id,
sum(case when product_id = 1 then 1 else 0 end) as prod_1_count,
sum(case when product_id = 2 then 1 else 0 end) as prod_2_count,
sum(case when product_id = 3 then 1 else 0 end) as prod_3_count,
sum(case when product_id = 4 then 1 else 0 end) as prod_4_count
from your_table
group by user_id
Upvotes: 108
Reputation: 2365
See if this works:
SELECT a.user_id,
(SELECT count(b.product_id) FROM sales b WHERE b.product_id = 1 AND a.user_id = b.user_id) AS prod_1_count,
(SELECT count(b.product_id) FROM sales b WHERE b.product_id = 2 AND a.user_id = b.user_id) AS prod_2_count,
(SELECT count(b.product_id) FROM sales b WHERE b.product_id = 3 AND a.user_id = b.user_id) AS prod_3_count,
(SELECT count(b.product_id) FROM sales b WHERE b.product_id = 4 AND a.user_id = b.user_id) AS prod_4_count
FROM sales a GROUP BY a.user_id;
Cheers. n.b. there may be slightly nicer ways to achieve the equivalent result.
Upvotes: 3