SQL Data Grouping - sql

Question

I was hoping someone would be able to help me (or point me in the right direction) on the following problem. I'm looking at grouping a large number of codes to only 3 digits, ensuring that if a participant had the code 122.2 and 122.3, it would count as one occurrence and not two.
Data Example:
Participant | group_code
1 | 1223
1 | 1224
1 | 1123
2 | 1012
2 | 0123
Current Code:
SELECT (left(group_code, 3)) as Group, count(left(group_code, 3)) as occurrence
from testDB
group by left(group_code, 3)
I suspect I need to use a unique element on the participant ID when grouping, however I'm not too sure.
Current Outcome:
Using the current data example, the result is as follows.
122 has 2 occurrences
112 has 1 occurrence
101 has 1 occurrence
012 has 1 occurrence
Expected Outcome:
122 has 1 occurrences
112 has 1 occurrence
101 has 1 occurrence
012 has 1 occurrence
Question: Is it possible to change the current code so that, if a single participant has multiple occurrences of a 3 digit value, for example 111.1, 111.2, 111.3, and 111.4, using the code above would provide the out 111 has occurred 4 times. However, I only want it to state it has appeared once as I'm only interested in a 3 digit level (and not the 4th).
Many thanks

Answer 1

Try this.
declare #t table ( group_code varchar(15))
insert into #t values ('122.2') ,('122.3' ) ,( '122.4' ) ,( '112.6'),( '112.0') , ( '119.1')
SELECT (left(group_code, 3)) as Grop,
count(left(group_code, 3)) as occurrence
from #t
group by left(group_code, 3)
select * from
(
SELECT (left(group_code, 3)) as Grop,
count(left(group_code, 3)) as occurrence
from #t
group by left(group_code, 3)
) a
join #t t on a.Grop = left(t.group_code, 3)

Answer 2

Create Table #T(Id int, Value decimal(16,2))
Insert into #T
Values(1,122.2),(1,122.3),(2,122.2)
Select Id,ROUND(Value,0)
from #T
Group By Id,ROUND(Value,0)

Answer 3

Try this
SELECT (left(group_code, 3)) as Group, count(*) as occurrence
from testDB
group by left(group_code, 3)

Answer 4

EDIT:
According to your last edit in your question it should be this:
DECLARE #tbl TABLE(Participant INT, group_code INT);
INSERT INTO #tbl VALUES
(1,1223)
,(1,1224)
,(1,1123)
,(2,1012)
,(3,0123);
WITH WithGroupingNew AS
(
SELECT tbl.*
,LEFT(CAST(tbl.group_code AS varchar(100)),3) AS NewGroupingCode
FROM #tbl AS tbl
)
,Counted AS
(
SELECT *
,ROW_NUMBER() OVER(PARTITION BY NewGroupingCode ORDER BY group_code) AS Counter
FROM WithGroupingNew
)
SELECT *
FROM Counted
WHERE Counter=1
This as before the edit, might be useful though...
With this CTE you get a field with the information you need:
DECLARE #tbl TABLE(Participant INT, group_code INT);
INSERT INTO #tbl VALUES
(1,1223)
,(1,1224)
,(1,1123)
,(2,1012)
,(3,0123);
WITH WithGroupingNew AS
(
SELECT tbl.*
,LEFT(CAST(tbl.group_code AS varchar(100)),3) AS NewGroupingCode
FROM #tbl AS tbl
)
SELECT * FROM WithGroupingNew
Result
Participant group_code NewGroupingCode
1 1223 122
1 1224 122
1 1123 112
2 1012 101
3 0123 123
It's on you to decide, what you are going to do with this:
SELECT DISTINCT NewGroupingCode FROM WithGroupingNew
or maybe
SELECT DISTINCT Participant,NewGroupingCode FROM WithGroupingNew
or maybe
SELECT *,ROW_NUMBER() OVER(PARTITION BY NewGroupingCode ORDER BY group_code)