avg of count incorrect results - sql

Question

I need to get the Brand and Type of all consoles that get repaired less than average, console types that haven't been repaired have to count towards this average, as well.
So I get the brand and type from the
console table.
And I join this to the Items table(artikel table in this DB).
Then I left join the items to the Repairs table,because I also need the console types that haven't been repaired, not sure if this is correct.
So now to get the amount of repairs per console type I did a count on the repaired_items_id in the Repairs table (repareerd_artikel_id in the picture),and I grouped it by the same column, and then I took the average of that count.
This is my syntax, I also tried different group by combinations but the results are always wrong.
select merk,type from console c join artikel a on
a.CONSOLE_ID=c.CONSOLE_ID left join REPARATIE r on
REPAREERD_ARTIKEL_ID=a.ARTIKEL_ID group by MERK,TYPE
HAVING (select avg(A.rcount) from (select
count(repareerd_artikel_id) AS rcount from REPARATIE group by
REPAREERD_ARTIKEL_ID) A) < (select avg(A.rcount) from (select
count(repareerd_artikel_id) AS rcount from REPARATIE group by
REPAREERD_ARTIKEL_ID) A)
And then I also tried starting with a count instead.
HAVING count(repareerd_artikel_id)< (select avg(A.rcount) from
(select count(repareerd_artikel_id) AS rcount from REPARATIE group by
REPAREERD_ARTIKEL_ID) A)
I have no idea what to do anymore now so any help would be much appreciated.

Answer 1

Using your query (and tbh not checking your images, as they are a pain to navigate to from SO) I think you could do something like this, breaking down the problem into three stages:
Work out the number of repairs for each console, including zeroes;
Work out the average number of repairs across all consoles;
List any repairs that are under the average.
WITH AllRepairs AS (
SELECT
merk,
[type],
ISNULL(COUNT(r.repareerd_artikel_id), 0) AS repairs
FROM
console c
INNER JOIN artikel a ON a.CONSOLE_ID = c.CONSOLE_ID
LEFT JOIN REPARATIE r ON r.REPAREERD_ARTIKEL_ID = a.ARTIKEL_ID
GROUP BY
merk,
[type]),
AverageRepairs AS (
SELECT
AVG(repairs) AS average_repairs
FROM
AllRepairs)
SELECT
a.*
FROM
AllRepairs a
CROSS JOIN AverageRepairs ar
WHERE
a.repairs < ar.average_repairs
ORDER BY
a.repairs;
You might want to worry about comparing integers to a decimal? For example, if your average number of repairs is 2.9 then only anything under 2 would count as below average. I think that's probably what you want?

Answer 2

The way I am viewing your info it looks like you do not have any repairs for the items in your items table. So, I added 2 repairs to one item and 1 repair to another. In my query, I produce an average across all 11 Merk/Types of 1.5. I then compare this to the number of repairs for each. The record with 2 repairs get presented in the result.
Create Table #console
(
Console_Id Int,
Merk Varchar(25),
Type VarChar(25),
Kleur VarChar(10),
Jaar_Uitgave Int,
Maat VarChar(10)
)
Insert Into #console Values
(1,'Sony 1','PS4 Slim','Wit',2016,'Slim'),
(2,'Microsoft','XBox','Beige',2004,'Port'),
(3,'Microsoft','XBox 360','Zwart',2011,'Pro'),
(4,'Microsoft','XBox One','Wit',2014,'Pro'),
(5,'Microsoft','XBox One X','Wit',2017,'Pro'),
(6,'Nintendo','NES Classic Edition','Wit',2016,'XL'),
(7,'Nintendo','Switch','Wit',2017,'XL'),
(8,'Nintendo','WII','Wit',2011,'Slim'),
(9,'Nintendo','WII Mini','Wit',2015,'XL'),
(10,'Nintendo','WII U','Wit',2013,'Slim'),
(11,'Sony','PS3','Wit',2013,'Port')
Create Table #Items
(
Artikel_ID Int,
BarCode VarChar(20),
Prijs Float,
Prijs_Per_D Float,
Spel_Of_Console VarChar(25),
Spel_ID Int,
Console_Id Int
)
Insert Into #Items Values
(301,'10000008',300.00,11.00,'Console',Null,3),
(302,'10000017',400.00,15.00,'Console',Null,4),
(303,'10000026',270.00,9.00,'Console',Null,9),
(304,'10000035',200.00,5.00,'Console',Null,6),
(305,'10000044',200.00,5.00,'Console',Null,11),
(306,'10000053',300.00,11.00,'Console',Null,12),
(307,'10000023',60.00,2.00,'Spel',15,Null),
(308,'10000242',36.00,2.00,'Spel',16,Null),
(309,'10000278',35.00,2.00,'Spel',21,Null),
(310,'10000107',66.00,4.00,'Spel',36,Null),
(311,'10000215',45.00,3.00,'Spel',40,Null)
Create Table #Repairs
(
Medewerker_Id Int,
Repareerd_Artikel_Id Int,
Schadenummer Int,
Huurovereenkomst_Id Int,
datum_Gereed DateTime,
Kosten Float,
Reparatiestatus VarChar(25)
)
Insert Into #Repairs Values
(1,259,7,12,'2017-08-03 00:00:00',112.00,'GEREED'),
(2,260,9,14,'2016-09-29 00:00:00',84.00,'GEREED'),
(3,288,19,28,'2017-04-09 00:00:00',96.00,'GEREED'),
(4,292,21,30,'2018-01-27 00:00:00',110.00,'GEREED'),
(5,283,16,24,'2015-12-29 00:00:00',103.00,'GEREED'),
(6,245,1,2,'2017-01-31 00:00:00',160.00,'GEREED'),
(7,245,2,3,'2018-01-18 00:00:00',120.00,'GEREED'),
(8,275,11,19,'2016-04-15 00:00:00',75.00,'GEREED'),
(9,276,12,20,'2015-08-25 00:00:00',174.00,'GEREED'),
(10,283,15,23,'2014-06-10 00:00:00',74.00,'GEREED'),
(11,297,21,34,'2014-07-17 00:00:00',96.00,'GEREED')
Insert Into #Repairs Values
(14,305,21,34,'2014-07-25 00:00:00',96.00,'GEREED'),
(12,301,21,34,'2014-07-17 00:00:00',96.00,'GEREED'),
(13,301,21,34,'2014-07-25 00:00:00',96.00,'GEREED')
Query
;With cte As
(
select
c.Merk, c.Type,
Count(r.REPAREERD_ARTIKEL_ID) As cnt
from
#console c Left join
#Items a on a.CONSOLE_ID=c.CONSOLE_ID left join
#Repairs r on r.REPAREERD_ARTIKEL_ID=a.ARTIKEL_ID
group by
c.merk, c.type
)
Select
*,
(Select Count(*) As totrecs From cte) As cntRecs ,
(Select avg(Cast(cte.cnt As Float)) As avgrecs From cte Where cte.cnt > 0) as avgrecs
From
cte
Where cte.cnt > (Select avg(Cast(cte.cnt As Float)) As avgrecs From cte Where cte.cnt > 0)
Result:
Merk Type cnt cntRecs avgrecs
Microsoft XBox 360 2 11 1.5