Question

I have a table Orders with order_date and order_total. Is it possible to find out how much was averaged ordered per month?

=== ORDER_DATE === ORDER_TOTAL
--- 7-JAN-10   --- 5271
--- 12-JAN-10  --- 3646
--- 15-JAN-10  --- 310
--- 5-FEB-10   --- 71173
--- 12-FEB-10  --- 45175
--- 25-FEB-10  --- 126

In this case it would be:

JAN - 3075.66667 // calculation: (5271 + 3646 + 310) / 3
FEB - 38824.6667 // calculation: (71173 + 45175 + 126) / 3

Answer 1

You’ll want to reduce each date to its month. The TRUNC() function below will return the first day of the month for each date. AVG() is an aggregate function, so you’ll want to GROUP BY the month.

SELECT TRUNC(order_date, 'MM') AS order_month, AVG(order_total) AS order_avg
  FROM orders
 GROUP BY TRUNC(order_date, 'MM')

Another way to accomplish this with analytic/window functions (this way you can return all the order data along with the average per month) would be as follows:

SELECT order_date, order_total, AVG(order_total) OVER (PARTITION BY TRUNC(order_date, 'MM'))
  FROM orders

The OP asked the following:

In the order order_month column I get something like 01-JAN-10 or
01-FEB-10. Is it also possible, to change it to FEB-10? Just without
the day.

I had suggested using TO_CHAR() with a mask of ‘MON-YY’ in this case but this meant that he could no longer sort properly on that column since it was now a string and not a date. One way to handle this would be as follows:

SELECT TO_CHAR(order_month, 'MON-YY'), order_avg
  FROM (
    SELECT TRUNC(order_date, 'MM') AS order_month, AVG(order_total) AS order_avg
      FROM orders
     GROUP BY TRUNC(order_date, 'MM')
     ORDER BY order_month
 )