VikingOnline
asked on
SQL simple select plus average of other lines
Hey guys,
I'm wondering how I can do something directly in the SQL statement without having to do two SQL statements and then have to loop through in code and match it together.
I want to select a list of all open orders, the name of the customer, the amount on the order - and the difficult part (for me) - the average order amount for that customer for orders that are closed.
So in mock up language it becomes something like:
SELECT [OrderNumber], [Customer], [Amount], (AVG([Amount] where Customer = same as this line and [status] = 'closed') AS "Avg Closed Orders Amount"
FROM tblOrders
WHERE [status] = 'open'
Is this possible to do directly in the SQL statement?
Or do I need to first pull all open orders. Then pull the average for closed orders grouped by customer, and then in programming code combine the two?
I can see that there is one very big difference between the two parts of the query. The first part is list per order number. (No group by or select distinct.) The average is grouped by customer, and if a customer has 5 open orders the same average will show up 5 times. And I think that's a problem, but I'm not sure if there are any ways around it.
I'm wondering how I can do something directly in the SQL statement without having to do two SQL statements and then have to loop through in code and match it together.
I want to select a list of all open orders, the name of the customer, the amount on the order - and the difficult part (for me) - the average order amount for that customer for orders that are closed.
So in mock up language it becomes something like:
SELECT [OrderNumber], [Customer], [Amount], (AVG([Amount] where Customer = same as this line and [status] = 'closed') AS "Avg Closed Orders Amount"
FROM tblOrders
WHERE [status] = 'open'
Is this possible to do directly in the SQL statement?
Or do I need to first pull all open orders. Then pull the average for closed orders grouped by customer, and then in programming code combine the two?
I can see that there is one very big difference between the two parts of the query. The first part is list per order number. (No group by or select distinct.) The average is grouped by customer, and if a customer has 5 open orders the same average will show up 5 times. And I think that's a problem, but I'm not sure if there are any ways around it.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
that style of correlated subquery (inside the selection clause) will be OK for small selections, but it won't perform very well for bigger selections. You could try this style of approach as an alternative
SELECT
A.[OrderNumber]
, A.[Customer]
, A.[Amount]
, B.avg_clord_amt AS "Avg Closed Orders Amount"
FROM tblOrders A
LEFT JOIN (
SELECT [Customer], AVG(Amount) avg_clord_amt
FROM tblOrders
WHERE STATUS = 'Closed'
GROUP BY Customer
) AS B ON A.Customer = B.Customer
WHERE A.[status] = 'open'
ASKER
So subquery was my initial thought and made sense to me.
I never thought about linking the table to itself.
How big would the data set be before you would see issues with subqueries?
(In this case, there can be a lot of orders - even if I put a date range on it.)
Sorry, I already awarded the points. Not sure if I can get that changed and split it, since it seems to be two good solutions to this.
I never thought about linking the table to itself.
How big would the data set be before you would see issues with subqueries?
(In this case, there can be a lot of orders - even if I put a date range on it.)
Sorry, I already awarded the points. Not sure if I can get that changed and split it, since it seems to be two good solutions to this.
both approaches use subqueries :)
and if doing date range selection don't use 'between'
see: "Beware of Between"
regarding points that's up to you, I'm fine as it is, but as a fyi you can ask for questions to be re-opened I believe, or simply ask another question.
The answer provided was fine for the question asked
select
...
, (select x from y where a.id = y.id) /* correlated, in the selection list */
select
...
from a
left join ( select ...) y on a.id = y.id
hard to put a number on when the lower approach will outperform the upper, but it most probably will. If you are doing a date range selection think about using the same or similar selection inside the group by too....
, (select x from y where a.id = y.id) /* correlated, in the selection list */
select
...
from a
left join ( select ...) y on a.id = y.id
and if doing date range selection don't use 'between'
see: "Beware of Between"
regarding points that's up to you, I'm fine as it is, but as a fyi you can ask for questions to be re-opened I believe, or simply ask another question.
The answer provided was fine for the question asked
ASKER
I can't believe how close I got in some of my attempts before giving up by the entire needing to GROUP BY, not thinking that I could just group by all levels of the details.
Thank you for the quick reply.