My scenario started off similar to a Island and Gaps problem, where I needed to find consecutive days of work. My current SQL query answers "ProductA was produced at LocationA from DateA through DateB, totaling X quantity".
However, this does not suffice when I needed to throw prices into the mix. Prices are in a separate table and handled in C# after the fact. Price changes are essentially a list of records that say "ProductA from LocationA is now Y value per unit effective DateC".
The end result is it works as long as the island does not overlap with a price-change date, but if it does overlap, I get a "close" answer, but it's not precise.
The C# code can handle applying the prices efficiently, what I need to do though is split the islands based on price changes. My goal is to make the SQL's partioning take into account the ranking of days from the other table, but I'm having trouble applying what I want to do.
The current SQL that generates my island is as follows
SELECT MIN(ScheduledDate) as StartDate, MAX(ScheduledDate) as
EndDate, ProductId, DestinationId, SUM(Quantity) as TotalQuantity
FROM (
SELECT ScheduledDate, DestinationId, ProductId, PartitionGroup = DATEADD(DAY ,-1 * DENSE_RANK() OVER (ORDER BY ScheduledDate), ScheduledDate), Quantity
FROM History
) tmp
GROUP BY PartitionGroup, DestinationId, ProductId;
The current SQL that takes from the PriceChange table and ranks the dates is as follows
DECLARE @PriceChangeDates TABLE(Rank int, SplitDate Date);
INSERT INTO @PriceChangeDates
SELECT DENSE_RANK() over (ORDER BY EffectiveDate) as Rank, EffectiveDate as SplitDate
FROM ProductPriceChange
GROUP BY EffectiveDate;
My thought is to somehow update the first queries inner SELECT statement to somehow take advantage of the @PriceChangeDates table created by the second query. I would think we can multiply the DATEADD's increment parameter by the rank from the declared table, but I am struggling to write it.
If I was to somehow do this with loops, my thought process would be to determine which rank the ScheduledDate would be from the @PriceChangeDates table, where its rank is the rank of the closest Date that is smaller than itself it can find. Then take whatever rank that gives and, I would think, multiply it by the increment parameter being passed in (or some math, for example doing a *@PriceChangeDates.Count() on the existing parameter and then adding in the new rank to avoid collisions). However, that's "loop" logic not "set" logic, and in SQL I need to think in sets.
Any and all help/advice is greatly appreciated. Thank you :)
UPDATE:
Sample data & example on SQLFiddle: http://www.sqlfiddle.com/#!18/af568/1
Where the data is:
CREATE TABLE History
(
ProductId int,
DestinationId int,
ScheduledDate date,
Quantity float
);
INSERT INTO History (ProductId, DestinationId, ScheduledDate, Quantity)
VALUES
(0, 1000, '20180401', 5),
(0, 1000, '20180402', 10),
(0, 1000, '20180403', 7),
(3, 5000, '20180507', 15),
(3, 5000, '20180508', 23),
(3, 5000, '20180509', 52),
(3, 5000, '20180510', 12),
(3, 5000, '20180511', 14);
CREATE TABLE PriceChange
(
ProductId int,
DestinationId int,
EffectiveDate date,
Price float
);
INSERT INTO PriceChange (ProductId, DestinationId, EffectiveDate, Price)
VALUES
(0, 1000, '20180201', 1),
(0, 1000, '20180402', 2),
(3, 5000, '20180101', 5),
(3, 5000, '20180510', 20);
The desired results would be to have a SQL statement that generates the result:
StartDate EndDate ProductId DestinationId TotalQuantity
2018-04-01 2018-04-01 0 1000 5
2018-04-02 2018-04-03 0 1000 17
2018-05-07 2018-05-09 3 5000 90
2018-05-10 2018-05-11 3 5000 26
To clarify, the end result does need the TotalQuantity of each split amount, so the procedural code that manipulates the results and applies the pricing knows how much of each product was one on each side of the price change to accurately determine the values.
