EXPERT RESPONSE
The problem, as stated, doesn't really make sense,
because the given query will return only one row, containing
the overall sum for the entire table.
So I will change it slightly. Let's say we're talking about orders placed
against items in an online web store. Each order can be
for one or more items, and within an order, the various
items all have sales amounts.
Now let's sum up the sales amounts of all the items
on each individual order.
select order_id
, sum(amount) as sumamount
from mytable
group
by order_id
So we might have one order for $9.37, two orders for $12.34,
one order for $23.45, four orders for $45.67, and so on. In order to
obtain this analysis of how many orders were for each different sumamount,
we'll use a derived table
select sumamount
, count(*) as occurrences
from (
select order_id
, sum(amount) as sumamount
from mytable
group
by order_id
) as dt
group
by sumamount
Note that some databases like Microsoft SQL Server
will require that you actually name the derived table,
so I always give it the name dt.
Naming the derived table is also necessary
if you wish to join it to other tables in the query.
|
