Postgres SQL calculate average of calculated totals

Question

I'm trying to write a query to return all products with higher than average dollar sales. The relevant tables and columns are:

• PRODUCTS: prod_id: integer, price: decimal
• ORDERLINES: prod_id: integer, quantity: integer

I can write the subquery that gets dollar sales for each product, but I'm having trouble applying the average function to that subquery. Everything I try returns a syntax error. Following is what I thought should be close to the answer:

select avg(sum) as avg_sales 
from (
   select sum(b.quantity * a.price) as total_sales 
   from products a, orderlines b 
   where a.prod_id = b.prod_id 
   group by a.prod_id
);

When I try this, I get:

ERROR: subquery in FROM must have an alias.

I don't get this, but I'm new to Postgres. Anyway, adding an alias gives me:

select avg(totals) as avg_sales 
from (
    select sum(b.quantity * a.price) as total_sales 
    from products a, orderlines b 
    where a.prod_id = b.prod_id group by a.prod_id
) as totals;

When I try this, I get: ERROR: function avg(record) does not exist

I realize the above code is just getting the overall average sales. If I can get average sales, then it should be easy to get the products that are above average.

Answer 1

When I try this, I get: ERROR: function avg(record) does not exist

This is because you are passing the alias of the derived table to the avg() function, which means you are passing a complete row (=record) to it, not a single column (value).

To get that working you need to use:

select avg(totals.total_sales) as avg_sales 
from (
    select sum(ol.quantity * p.price) as total_sales 
    from products p 
      join orderlines ol on p.prod_id = ol.prod_id
    group by p.prod_id
) as totals;

Note that I replaced the ancient, outdated and fragile implicit join with a modern explicit JOIN operator. If you are learning SQL you should get used to that syntax.

---

To get the products with a higher then average sale, you need to calculate both: the sales per product and the average sale.

This can be achieved by using a window function while aggregating:

select p.prod_id, 
       sum(ol.quantity * p.price) as total_sales, 
       avg(sum(ol.quantity * p.price)) over () as average_sales
from products p 
  join orderlines ol on p.prod_id = ol.prod_id
group by p.prod_id;

This can now be wrapped in a derived table to filter out those with a lower total sales:

select *
from (
  select p.prod_id, 
         sum(ol.quantity * p.price) as total_sales, 
         avg(sum(ol.quantity * p.price)) over () as average_sales
  from products p 
    join orderlines ol on p.prod_id = ol.prod_id
  group by p.prod_id
) t
where total_sales >= average_sales;

SQLFiddle example: http://sqlfiddle.com/#!15/7f8ab/1

Answer 2

You can try something like this. First using a sub query find the average sales, then get all the products above that sales

SELECT p.prod_id, o.quantity
    FROM products p INNER JOIN orderlines o
    ON p.prod_id = o.prod_id
    WHERE o.quantity  >  (      SELECT AVG(ol.quantity)
                                FROM products pr INNER JOIN orderlines ol
                                ON pr.prod_id = ol.prod_id
                        )