sql - salesmen by territory query - combine multiple statements with WITH clause

Question

We have a territory/salesman relationship where: each Territory (TERR) may have multiple salesman (SALM), but each salesman can have only one territory.

Example:

Territory      Salesman   Sales$ Period   
    1             100      999     1
    1             100      999     2
    1             200      999     1 
    1             200      999     2    
    2             300      999     1
    2             300      999     2
    3             400      999     1
    3             400      999     2

I would like to build a summary table, preferably using a single SQL statement that -

• shows territory,
• number of salesmen for this territory (** CLARIFICATION: ** that is total unique salesmen codes that made sales in this territory) and
• list the salesmen numbers in the same file.

example:

Territory # of Salesmen   Salesman   
--------- -------------   --------   
1             2            100       
1             2            200       
2             1            300       
3             1            400       

SQL:

// First I get a list of territories and number of salesman in each territory
With tmpFile as (Select TERR, SALM, count(*) 
   FROM FILE
   GROUP BY TERR, SALM 
   ORDER BY TERR, SALM   
) 
SELECT tmpFile.TERR, count(*) FROM tmpFile GROUP BY          
tmpFile.TERR ORDER BY tmpFile.TERR

// Next step is to get a list of all salesmen in a territory
Select TERR, SALM
   FROM FILE
   GROUP BY TERR, SALM 
   ORDER BY TERR, SALM
) 

// Final step would be to combine the above two steps. I would like to combine all three steps into a single statement if possible. Is there more effective way to do this. I am using IBM db/2 SQL.

Answer 1

While I would like to use the OLAP version of the statement, my version of DB2 (V6R1) doesn't support the COUNT() function in that context (boohoo). That said, here's a version that does work, and should work on any other RDBMS (I think):

SELECT DISTINCT terr, salm, salesmen
FROM File -- I really hope that's not your file name - it's a reserved word
JOIN (SELECT terr territory, COUNT(DISTINCT salm) salesmen
      FROM File
      GROUP BY terr) Salesmen_Per_Territory
ON territory = terr

Please note the DISTINCT inside COUNT(), as this is what is keeping the salesmen from being counted for each period.

Answer 2

Why not do it in single query?

SELECT Territory, SalesMan, COUNT(*) as [No. of Salesman]
FROM tableName
GROUP BY Territory, SalesMan

Are they on different table? eg, TERR and SALM?

SELECT  a.Territory, b.SalesMan, Count(b.Territory) as [No. of Salesman]
FROM    Terr a 
            INNER JOIN SALM b
                on a.Territory = b.Territory
GROUP BY a.Territory, b.SalesMan

Answer 3

I'd say you are on the right track using a common table expression [aka. WITH clause], but it can easily and efficiently be done in one statement.

WITH t as
(SELECT terr, count(distinct salm) as slspeople
   FROM yourfile
   WHERE salesamt > 0
   GROUP BY terr
)
SELECT DISTINCT s.terr, t.slspeople, s.salm
  FROM yourfile as s
  JOIN    t       using (terr)
  WHERE salesamt > 0
  ORDER BY s.terr, s.salm

The using (terr) clause is a shorthand for on s.terr = t.terr, where the matching columns must have the same name in both tables.

I added the WHERE salesamt > 0 only because one of your clarification requirements was to only list salespeople who had sales. If this is not applicable take this out, of course.

---

It may be a good idea to take the various answers provided, and analyze them with Visual Explain. This is a valuable tool where you can try different variations, and see what performs better. I use this on IBM i, but I believe it exists on the other two platforms as well.