I have a table that contains sales data with the following columns that are important to me: Customer#, SalesDate,...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
SalesAmount. I would like to be able to total the SalesAmount into yearly quarters (i.e. Q1, Q2,...), for each customer. Is it possible to do this in one query? I'm using a Progress database, so any functions available in Access/SQLServer/Oracle are not available to me, I'm afraid. Thank you!
Allow me to offer my condolences. A quick Web search turns up two references: Progress SQL-89 Guide and Reference and Progress SQL-92 Guide and Reference. Sure enough, the SQL-92 version supports a substantial list of scalar functions including DAYOFYEAR, so you could write:
select Customer# , case dayofyear(SalesDate) when < 92 then 1 when < 183 then 2 when < 274 then 3 else 4 end as Quarter , sum(SalesAmount) from SalesData group by Customer# , case dayofyear(SalesDate) when < 92 then 1 when < 183 then 2 when < 274 then 3 else 4 end
However, if you are using the SQL-89 version, it looks like you need another way to find the day number. There is no indication that you could do something even as simple as adding an integer to a date (see The integers table for an example of generating a series of dates). The only obvious solution that requires no date calculations or functions is a calendar table which will allow you to associate every date with its quarter:
create table Calendar ( CalDate date not null , Q smallint not null ). insert into Calendar values ('01/01/04',1). insert into Calendar values ('01/02/04',1). insert into Calendar values ('01/03/04',1). insert into Calendar values ('01/04/04',1). ... insert into Calendar values ('12/29/04',4). insert into Calendar values ('12/30/04',4). insert into Calendar values ('12/31/04',4).
Two quirks that I spotted in the SQL-89 manual: each statement must end in a period, and date strings must be specified as mm/dd/yy. Now you can say:
select Customer# , Q , sum(SalesAmount) from Calendar , SalesData where CalDate = SalesDate group by Customer# , Q.
One final note: in the first solution above, the assumption is that the first 91 days of the calendar year are in Q1, the second 91 in Q2, and the last 92 or 93, as the case may be, are in Q4. Realistically, this just doesn't happen.
In most organizations that use quarters or similar types of fiscal period, everything is measured in whole weeks. In 2004, January 1 through 3 would most likely be part of Q4 of 2003, since there are more 2003 days in that week than 2004 days. Likewise, January 1st of 2005 would likely be part of Q4 in 2004. When your quarters are actually fiscal quarters, you really do need to use a calendar table, and it should look like this:
create table Calendar ( CalDate date not null , FY smallint not null , FQ smallint not null ). insert into Calendar values ('01/01/04',2003,4). insert into Calendar values ('01/02/04',2003,4). insert into Calendar values ('01/03/04',2003,4). insert into Calendar values ('01/04/04',2004,1). ... insert into Calendar values ('12/29/04',2004,4). insert into Calendar values ('12/30/04',2004,4). insert into Calendar values ('12/31/04',2004,4). insert into Calendar values ('01/01/05',2004,4). insert into Calendar values ('01/02/05',2005,1).
Now just join on the date and group by the fiscal year and fiscal quarter.
Dig Deeper on Oracle and SQL
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.