Using LAG with Grouping Sets produces unexpected result

Question

The following query always sets ReturnWithin to 0 for the second Grouping Set and I don't understand why:

                    SELECT
                        CardNumberKey,
                        Months,
                        0 AS Rolling12MonthInd,
                        Months - LAG( Months,1 ) OVER( PARTITION BY CardNumberKey ORDER BY Months ) AS ReturnWithin,
                        [MonthName],
                        Site_ID,
                        Site_Name,
                        SUM( Transaction_Count ) AS Transaction_Count,
                        SUM( Transaction_Amount ) AS Transaction_Amount
                    FROM
                        #PreFirstStage
                    WHERE
                        CardNumberKey = 54072178
                    GROUP BY GROUPING SETS
                        (
                            (
                                CardNumberKey,
                                Months,
                                [MonthName],
                                Site_ID,
                                Site_Name
                            ),
                            (
                                CardNumberKey,
                                Months,
                                [MonthName]
                            )
                        )

I restricted the list to that single card as I knew it had only been used once for the account (all sites) so should have produced NULL for ReturnedWithin for both the Site level and overall level.

So, in the results from #PreFirstStage there is only 1 row for that Card Number Key. If I do an ordinary group by using that second lot of criteria on the same source data set then it produces the expected result.

It's the first time I've tried to use Grouping Sets so it is more than likely my lack of understanding of how it works.

So I'll probably just do it as 2 queries with a UNION ALL just to keep my project moving but I'd love to know how to make this work as I've written it above.

Answer 1

Hi,
Regarding the usage of GROUPING SETS, please refer to the following example:
Please refer to:

CREATE TABLE inventory (
    warehouse VARCHAR(255),
    product VARCHAR(255) NOT NULL,
    model VARCHAR(50) NOT NULL,
    quantity INT,
    PRIMARY KEY (warehouse,product,model)
);

INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Jose', 'iPhone','6s',100);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Fransisco', 'iPhone','6s',50);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Jose','iPhone','7',50);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Fransisco', 'iPhone','7',10);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Jose','iPhone','X',150);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Fransisco', 'iPhone','X',200);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Jose','Samsung','Galaxy S',200);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Fransisco','Samsung','Galaxy S',200);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Fransisco','Samsung','Note 8',100);
INSERT INTO inventory(warehouse, product, model, quantity)
VALUES('San Jose','Samsung','Note 8',150);


SELECT * FROM inventory;

SELECT warehouse,product, SUM (quantity) qty
FROM inventory
GROUP BY warehouse,product;

SELECT warehouse,SUM (quantity) qty
FROM inventory
GROUP BY warehouse;

SELECT product,SUM (quantity) qty
FROM inventory
GROUP BY product;

SELECT SUM(quantity) qty
FROM inventory;

SELECT warehouse,product, SUM (quantity) qty
FROM inventory
GROUP BY
    GROUPING SETS((warehouse,product),(warehouse),(product),());

For more details, please refer to:
SQL GROUPING SETS

If you have any question, please feel free to let me know.

Regards
Echo

---

If the answer is helpful, please click "Accept Answer" and upvote it.

Answer 2

You say

Months - LAG( Months,1 ) OVER( PARTITION BY CardNumberKey ORDER BY Months ) AS ReturnWithin

In your two-row table example, CardNumberKey and Months have the same value on both rows. Thus they are in the same partition. And 24328-24238 gives NULL.

If you want the partition to restart when you come to the higher aggregation level, you need to add something to the PARTITION BY clause. This worked on your small table:

Months - LAG( Months,1 ) OVER( PARTITION BY CardNumberKey, 
                                   GROUPING_ID(CardNumberKey, Months, MonthName, Site_ID, Site_Name) 
                             ORDER BY Months ) AS ReturnWithin,

But I don't use this function, and it's cousing GROUPING that often, so I am not sure that this is the exact solution.