Re: Need an SQL joining lesson -- MIDRANGE-L

Never told you the PK's

CUSRBTMST = CUSNR
DMCATMST = CTCODE
SLSALL = none (there are 6 indexes)
CRDMEDTL = none (there are 10 indexes)

All files are defined via DDL, not DDS.

These are the actual results I am expecting:

Cust YYYYMM Category Sales Credits
==== ====== ======== ======= =======
1892 201802 B 1133.29 .00
1892 201802 D 236.98 .00
1892 201802 E 1741.93 420.01
1892 201802 F 1770.11 137.76
4554 201802 B 3556.01 72.48
4554 201802 D 1004.97 7.12
4554 201802 E 2853.17 198.25
4554 201802 F 3081.09 .00
(Total) 15377.55 835.62

With this SQL:

SELECT m.cusnr, 201802 AS YYYYMM, COALESCE(s.catcd1, c.catcd1, ' ') AS
category,
COALESCE(SUM(s.itnsa), 0) AS Sales,COALESCE(SUM(c.cdtam), 0) AS Credits
FROM DBMSTF.CUSRBTMST m
CROSS JOIN DBMSTF.DMCATMST cat
LEFT JOIN SLSHST.SLSALL s ON m.cusnr = s.cusnr AND s.catcd1 = cat.ctcode
LEFT JOIN DBMSTF.CRDMEDTL c ON m.cusnr = c.cusnr AND c.catcd1 =
cat.ctcode
WHERE s.INVDT BETWEEN '2018-02-01' AND '2018-02-28' AND
c.DTWRTISO BETWEEN '2018-02-01' AND '2018-02-28'
GROUP BY m.CUSNR, COALESCE(s.catcd1, c.catcd1, ' ')
ORDER BY 1, 2

I get these results:

Cust YYYYMM Category Sales Credits
==== ====== ======== ======= =======
1892 201802 E 6967.72 8400.20
1892 201802 F 3540.22 5097.12
4554 201802 B 3556.01 3406.56
4554 201802 D 1004.97 163.76
4554 201802 E 5706.34 6344.00
(Total) 20775.26 23411.64

I note a few things:

1) more credit than sales, never a good thing from a business standpoint :)

2) numbers obviously not right

3) if there are sales and no credits, no record appears and I assume the
reverse would be true

On Thu, Apr 12, 2018 at 7:50 PM, Charles Wilt <charles.wilt@xxxxxxxxx>
wrote:

That's not going to work...

Consider
Customer
Cusnr
1

Sales
Cusnr Cat Amt
1 A 5.00
1 B 10.00

Credits
1 D -1.00
1 E -0.50

The intermediate table (before aggregation) will be
CUSNR CUSNR CAT AMT CUSNR CAT
AMT
1 1 A 5.00 1 D
1.00-
1 1 A 5.00 1 E
.50-
1 1 B 10.00 1 D
1.00-
1 1 B 10.00 1 E
.50-

basically cross joining between sales & credits since you're joining on
just customer...

We need to add a CATEGORY_MASTER

Then our basic statement ends up with
select *
from cust cst
cross join CATEGORY_MASTER cat
left outer join sales sls on cst.cusnr = sls.cusnr
and cat.catcd1 = sls.catcd1
left outer join credits crd on cst.cusnr = crd.cusnr
and cat.catcd1 = crd.catcd1

And our intermediate table looks like...
CUSNR CAT CUSNR CAT AMT CUSNR
CAT AMT
1 A 1 A 5.00 -
- -
1 B 1 B 10.00 -
- -
1 C - - - -
- -
1 D - - - 1
D 1.00-
1 E - - - 1
E .50-

Adding SUM and group by cst.cusnr, cat.catcd1
CUSNR CAT SALES_TOTAL
CREDIT_TOTAL
1 B 10.00
-
1 E -
.50-
1 D -
1.00-
1 C -
-
1 A 5.00
-

Lets add some more sales & credits for the existing cats...
insert into sales values (1,'A',15.00), (1,'B',12.00)
insert into credits values (1,'D',-0.75), (1,'E',-.25)

CUSNR CAT SALES_TOTAL
CREDIT_TOTAL
1 B 22.00
-
1 E -
.75-
1 D -
1.75-
1 C -
-
1 A 20.00
-

So yeah, could be done without a CTE...

Charles

On Thu, Apr 12, 2018 at 5:07 PM, Mark Murphy <jmarkmurphy@xxxxxxxxx>
wrote:

I think the CTE is overthinking it a bit here, How about this?

SELECT m.cusnr,
201802 AS YYYYMM,
COALESCE(s.catcd1, c.catcd1, ' ') AS category,
COALESCE(SUM(s.itnsa), 0) AS Sales,
COALESCE(SUM(c.cdtam), 0) AS Credits
FROM DBMSTF.CUSRBTMST m
left JOIN SLSHST.SLSALL s ON m.cusnr = s.cusnr
left JOIN DBMSTF.CRDMEDTL c ON m.cusnr = c.cusnr
WHERE s.INVDT BETWEEN '2018-02-01' AND '2018-02-28' AND
c.DTWRTISO BETWEEN '2018-02-01' AND '2018-02-28'
GROUP BY m.CUSNR, COALESCE(s.catcd1, c.catcd1, ' ')
ORDER BY 1,2,3

I think that by grouping by the categories in the individual files
seperately, you are getting every combination of record that could exist.
But the Groupby can contain an expression, and by using the value that

will

appear in the output, you should get the right output.

On Thu, Apr 12, 2018 at 6:42 PM, Charles Wilt <charles.wilt@xxxxxxxxx>
wrote:

You're running into a well known issue call "fan-out"

Here's a nice animated picture
https://discourse.looker.com/t/outer-join-on-false-or-how-
i-learned-to-stop-fanning-out-and-love-the-null/4786

The FULL OUT JOIN solution that page presents is a new one to me..I'm

going

to have to play with it. :)

How I'd normally solve the problem would be with CTEs that summarize

the

data first, then join the CTEs.

The key is that you want a single row for Customer, Category,
YearMonth...so you use CTEs to transform the raw data into that...
-- Note make sure DATFMT is *ISO
with sales as (
select cusnr, catcd1, substr(char(invdt),1,7) as YearMonth
sum(itnsa) as total
from slshst.slsall
where invdat between '2018-02-01' AND '2018-02-28'
group by cusnr, catcd1, substr(char(invdt),1,7)
), credits as (
select cusnr, catcd1, substr(char(dtwrtiso),1,7) as YearMonth
sum(cdtam) as total
from dbmstf.crdmedtl
where dtwrtiso between '2018-02-01' AND '2018-02-28'
group by cusnr, catcd1, substr(char(dtwrtiso),1,7)
), custs_cats as (
select cst.cusnr, cat.catcd1, '2018-02' as YearMonth
from dbmstf.cusrbtmst cst
cross join CATEGORY_MASTER cat
)
select cc.cusnr, cc.catcd1, cc.YearMonth
, coalesce(sls.Total, 0) as sales_total
, coalesce(crd.total, 0) as credit_total
from custs_cats cc
left outer join sales sls
on cc.cusnr = sls.cusnr
and cc.catcd1 = sls.catcd1
and cc.YearMonth = sls.YearMonth
left outer join credits crd
on cc.cusnr = crd.cusnr
and cc.catcd1 = crd.catcd1
and cc.YearMonth = crd.YearMonth;

Also note that a "Dates Table" or "Calendar Table", with a YearMonth

column

among many others comes in quite handy for things like this.

Lots of date related queries are become much easier with such a table.

Charles

On Thu, Apr 12, 2018 at 1:49 PM, Jeff Crosby <

jlcrosby@xxxxxxxxxxxxxxxx>

wrote:

Not the greatest at SQL but have a project where I think it's a

perfect

fit

so I want to figure this out. It's ultimately going into an RPG

program

(as an INSERT statement), but the question at the moment is the SQL,

not

RPG. That's why what you see below is a SELECT with stuff hardcoded.

I

just want the right numbers first.

Some customers are going to start getting rebates. The rebate will a
percent of (sales - credits) by item category for the month. I want

to

insert into a file the following 5 fields:

customer number
YYYYMM
category
total sales for the month
total credits for the month

There is a file of customers and their rebate percents, the sales

file

(with the category field in it), and a credits file (with the

category

field in it). There could be both sales and credits for a
customer/category, only sales for a customer/category, or only

credits

for

a customer/category. In all 3 cases I want 1 record per

customer/category.

CUSRBTMST - rebates percents file
SLSALL - sales file
CRDMEDTL - credits file

This is the SQL I have now:

SELECT m.cusnr,
201802 AS YYYYMM,
COALESCE(s.catcd1, c.catcd1, ' ') AS category,
COALESCE(SUM(s.itnsa), 0) AS Sales,
COALESCE(SUM(c.cdtam), 0) AS Credits
FROM DBMSTF.CUSRBTMST m
JOIN SLSHST.SLSALL s ON m.cusnr = s.cusnr
JOIN DBMSTF.CRDMEDTL c ON m.cusnr = c.cusnr
WHERE s.INVDT BETWEEN '2018-02-01' AND '2018-02-28' AND
c.DTWRTISO BETWEEN '2018-02-01' AND '2018-02-28'
GROUP BY m.CUSNR, s.catcd1, c.catcd1
ORDER BY m.CUSNR, s.catcd1, c.catcd1

That's a copy/paste from ACS run SQL scripts that I formatted a bit
further.

The results I get have no resemblance to the actual sales and

credits.

:)

I'm joining the rebate percent file to both the sales and credits

file.

That might be the problem, I don't know. Are 2 SELECTS required?

Thanks for any pointers.

--

Jeff Crosby
VP Information Systems
UniPro FoodService/Dilgard
P.O. Box 13369
Ft. Wayne, IN 46868-3369
260-422-7531
direct.dilgardfoods.com

The opinions expressed are my own and not necessarily the opinion of

my

company. Unless I say so.
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L)

mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

Help support midrange.com by shopping at amazon.com with our

affiliate

link: http://amzn.to/2dEadiD

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

Help support midrange.com by shopping at amazon.com with our affiliate
link: http://amzn.to/2dEadiD

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

Help support midrange.com by shopping at amazon.com with our affiliate
link: http://amzn.to/2dEadiD

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

Help support midrange.com by shopping at amazon.com with our affiliate
link: http://amzn.to/2dEadiD