I'm working a query and I wasn't sure if my current approach is the "best" approach.I got a table filled with values and a table with buckets that I use to group those values.
There are 2 types of buckets:
- A priority bucket, if items apply to the conditions they should fall in this bucket and be excluded for any following bucket.
- A simple range bucket, the values should fall within a range.
I set up a simplified version that can be used to test.the [Filter] column in #tempBUcket here is used for the priority buckets
The table definitions
create table #TempData([Id] int, [value] int);create table #TempBucket([Id] int, [FromVal] int,[ToVal] int,[Filter] int);Some test data
INSERT INTO #TempData (Id,value) VALUES (1,1),(2,2),(3,3),(4,4),(5,5), (6,6),(7,7),(8,8),(9,9),(10,1), (11,2),(12,3),(13,4),(14,5),(15,6), (16,7),(17,8),(18,9),(19,10),(20,11);INSERT INTO #TempBucket(Id,FromVal,ToVal,Filter) VALUES (1,0,3,0),(2,3,6,0), (3,6,9,0),(4,9,12,0), (5,NULL,NULL,1),(6,NULL,NULL,2);Right now I have this query which works and gives the results I want but I feel this could be done better and more efficiently.
WITH filteredData as (SELECT tv.*,COALESCE(tb1.Id,tb2.Id) as BucketId FROM #TempData tv LEFT JOIN #TempBucket tb1 --prio bucket join ON tb1.Filter IS NOT NULL AND tb1.Filter = tv.Value LEFT JOIN #TempBucket tb2 --range bucket join ON tb2.Fromval < tv.value AND tv.value <= tb2.ToVal )SELECT Count(fd.value) as count ,BucketId FROM filteredData fd GROUP BY BucketId;