Erroneous SUM when using JOIN on SQL request - sql

Question

Hi i have two tables with a lot of data. I have to get some data of both tables. to do this i use an inner join. But when i use SUM i get wrong result. i know that the result is multiplied as many rows returned. How can i get arround this ?
The request :
select SUM(CASE WHEN det.IdGarantie != 18 AND det.IdGarantie != 17 AND det.IdGarantie!= 11 THEN det.Q_PrimeNette ELSE 0 END) AS 'cot nette',
SUM(CASE WHEN det.IdGarantie = 18 THEN det.Q_PrimeNette ELSE 0 END) AS 'Fond comp',
SUM(CASE WHEN det.IdGarantie = 11 THEN det.Q_PrimeNette ELSE 0 END) AS 'Carte Verte',
SUM(CASE WHEN det.IdGarantie = 17 THEN det.Q_PrimeNette ELSE 0 END) AS 'Adhesion',
SUM(CASE WHEN det.IdGarantie != 18 AND det.IdGarantie != 17 AND det.IdGarantie!= 11 THEN det.Q_Taxe ELSE 0 END) AS 'Taxe',
SUM(CASE WHEN det.IdGarantie = 18 THEN det.Q_Taxe ELSE 0 END) AS 'Taxe Fond Comp',
SUM(CASE WHEN det.IdGarantie = 17 THEN det.Q_Taxe ELSE 0 END) AS 'Taxe Adhésion',
SUM(CASE WHEN det.IdGarantie = 11 THEN det.Q_Taxe ELSE 0 END) AS 'Taxe carte verte',
SUM(hst.Q_Access) as 'Access'
from OtoHistorique hst
inner Join OtoHistoriqueDet det On det.IdHistorique = hst.IdHistorique
where hst.POLICE = 3221086
i get 60 in Access column when the expected result is 20
Anyone can help me with this ? Sorry if i use bad syntax in my request but i'm just a beginner in SQL and for the moment the result is more important than the syntax.

Answer 1

You probably have one-to-many relationship between the tables, which means some of the values for SUM get duplicated.
You can go around this by avoiding the join like this:
select SUM(CASE WHEN det.IdGarantie != 18 AND det.IdGarantie != 17 AND det.IdGarantie!= 11 THEN det.Q_PrimeNette ELSE 0 END) AS 'cot nette',
SUM(CASE WHEN det.IdGarantie = 18 THEN det.Q_PrimeNette ELSE 0 END) AS 'Fond comp',
SUM(CASE WHEN det.IdGarantie = 11 THEN det.Q_PrimeNette ELSE 0 END) AS 'Carte Verte',
SUM(CASE WHEN det.IdGarantie = 17 THEN det.Q_PrimeNette ELSE 0 END) AS 'Adhesion',
SUM(CASE WHEN det.IdGarantie != 18 AND det.IdGarantie != 17 AND det.IdGarantie!= 11 THEN det.Q_Taxe ELSE 0 END) AS 'Taxe',
SUM(CASE WHEN det.IdGarantie = 18 THEN det.Q_Taxe ELSE 0 END) AS 'Taxe Fond Comp',
SUM(CASE WHEN det.IdGarantie = 17 THEN det.Q_Taxe ELSE 0 END) AS 'Taxe Adhésion',
SUM(CASE WHEN det.IdGarantie = 11 THEN det.Q_Taxe ELSE 0 END) AS 'Taxe carte verte',
(SELECT SUM(hst.Q_Access)
FROM OtoHistorique hst
WHERE det.IdHistorique = hst.IdHistorique) as 'Access'
FROM OtoHistoriqueDet det
WHERE EXISTS (SELECT 1
FROM OtoHistorique hst2
WHERE hst2.POLICE = 3221086
AND hst2.IdHistorique = det.IdHistorique)
This could probably be more optimized but we'd need more info on the logic between the two tables.

Answer 2

One approach is to do the aggregation before the join:
select det.*,
SUM(hst.Q_Access) as "Access"
from OtoHistorique hst inner Join
(select det.IdHistoric,
SUM(CASE WHEN det.IdGarantie != 18 AND det.IdGarantie != 17 AND det.IdGarantie!= 11 THEN det.Q_PrimeNette ELSE 0 END) as "cot nette"
SUM(CASE WHEN det.IdGarantie = 18 THEN det.Q_PrimeNette ELSE 0 END) AS "Fond comp",
SUM(CASE WHEN det.IdGarantie = 11 THEN det.Q_PrimeNette ELSE 0 END) AS "Carte Verte",
SUM(CASE WHEN det.IdGarantie = 17 THEN det.Q_PrimeNette ELSE 0 END) AS "Adhesion",
SUM(CASE WHEN det.IdGarantie != 18 AND det.IdGarantie != 17 AND det.IdGarantie!= 11 THEN det.Q_Taxe ELSE 0 END) AS "Taxe",
SUM(CASE WHEN det.IdGarantie = 18 THEN det.Q_Taxe ELSE 0 END) AS "Taxe Fond Comp",
SUM(CASE WHEN det.IdGarantie = 17 THEN det.Q_Taxe ELSE 0 END) AS "Taxe Adhésion",
SUM(CASE WHEN det.IdGarantie = 11 THEN det.Q_Taxe ELSE 0 END) AS "Taxe carte verte",
SUM(hst.Q_Access) as "Access"
from OtoHistoriqueDet det
group by det.IdHistorique
) det
On det.IdHistorique = hst.IdHistorique
where hst.POLICE = 3221086;
Because you are filtering the query, this might be expensive. The entire details table has to be aggregated. So, you can do the join in the subquery for filtering purposes and then again for the calculation:
select det.*,
SUM(hst.Q_Access) as "Access"
from OtoHistorique hst inner Join
(select det.IdHistoric,
SUM(CASE WHEN det.IdGarantie != 18 AND det.IdGarantie != 17 AND det.IdGarantie!= 11 THEN det.Q_PrimeNette ELSE 0 END) as "cot nette"
SUM(CASE WHEN det.IdGarantie = 18 THEN det.Q_PrimeNette ELSE 0 END) AS "Fond comp",
SUM(CASE WHEN det.IdGarantie = 11 THEN det.Q_PrimeNette ELSE 0 END) AS "Carte Verte",
SUM(CASE WHEN det.IdGarantie = 17 THEN det.Q_PrimeNette ELSE 0 END) AS "Adhesion",
SUM(CASE WHEN det.IdGarantie != 18 AND det.IdGarantie != 17 AND det.IdGarantie!= 11 THEN det.Q_Taxe ELSE 0 END) AS "Taxe",
SUM(CASE WHEN det.IdGarantie = 18 THEN det.Q_Taxe ELSE 0 END) AS "Taxe Fond Comp",
SUM(CASE WHEN det.IdGarantie = 17 THEN det.Q_Taxe ELSE 0 END) AS "Taxe Adhésion",
SUM(CASE WHEN det.IdGarantie = 11 THEN det.Q_Taxe ELSE 0 END) AS "Taxe carte verte"
from OtoHistoriqueDet det join
OtoHistorique hst
On det.IdHistorique = hst.IdHistorique and hst.POLICE = 3221086
group by det.IdHistorique
) det
On det.IdHistorique = hst.IdHistorique
where hst.POLICE = 3221086;