Question

I am trying to populate a single average based off a CASE statement utilizing a WHERE clause with MONTHS of a COUNT, however it is returning an invalid use of group function Error Code 1111:

SELECT 
count(ORDER_NUMBER) as cnt,
a.ITEM,
MONTH(TRANS_DATE) as Month,
AVG(CASE
WHEN count(ORDER_NUMBER) IS NULL THEN 0
WHEN count(ORDER_NUMBER) < 1 AND a.CAT != 'E' THEN 0
WHEN count(ORDER_NUMBER) < 1 AND a.CAT = 'E' THEN .05
WHEN count(ORDER_NUMBER) >= 1 AND count(ORDER_NUMBER) < 4 AND a.CAT = 'E' THEN .5
WHEN count(ORDER_NUMBER) >= 4 AND a.CAT = 'E' THEN 1
WHEN count(ORDER_NUMBER) <= 5 AND a.CAT != 'E' THEN .05
WHEN count(ORDER_NUMBER) >= 6 AND count(ORDER_NUMBER) < 11 AND a.CAT != 'E' THEN .25
WHEN count(ORDER_NUMBER) >= 11 AND count(ORDER_NUMBER) < 21 AND a.CAT != 'E' THEN .5
WHEN count(ORDER_NUMBER) >= 21 AND a.CAT != 'E' THEN 1 END) as TSCORE
FROM trx.TRX_Import t2
    JOIN ATTRIBUTES a
    ON a.ITEM = t2.ITEM
WHERE `TRANS_DATE` BETWEEN DATE_SUB(CURDATE(), INTERVAL 365 DAY) AND CURDATE()
AND t2.ITEM = 'WARI-BB150'

When I add the additional GROUP BY MONTH It correctly identifies each months TSCORE…I would rather it just spit out one AVERAGE of all those TSCORES.

With GROUP BY MONTH, it spits out:

   cnt     ITEM        Month   TSCORE
    1   WARI-BB150      1       0.5
    2   WARI-BB150      2       0.5
    1   WARI-BB150      3       0.5
    2   WARI-BB150      4       0.5
    2   WARI-BB150      5       0.5
    2   WARI-BB150      6       0.5
    4   WARI-BB150      7       1
    2   WARI-BB150      10      0.5
    3   WARI-BB150      11      0.5
    1   WARI-BB150      12      0.5

I would prefer the data to spit out just the average TSCORE of an item based of the above table (so only 1 row of info):

cnt   ITEM          TSCORE
20    WARI-BB150     0.55

Answer 1

I had to use a subquery in the FROM clause and create an AVG on that:

SELECT
t2.ITEM as Item,
AVG(t2.TSCORE) as AvgTSCORE
FROM 
(SELECT count(ORDER_NUMBER) as cnt, t3.ITEM, MONTH(TRANS_DATE) as Month, 
(CASE
WHEN count(ORDER_NUMBER) IS NULL THEN 0
WHEN count(ORDER_NUMBER) < 1 AND a.CAT != 'E' THEN 0
WHEN count(ORDER_NUMBER) < 1 AND a.CAT = 'E' THEN .05
WHEN count(ORDER_NUMBER) >= 1 AND count(ORDER_NUMBER) < 4 AND a.CAT = 'E' THEN .5
WHEN count(ORDER_NUMBER) >= 4 AND a.CAT = 'E' THEN 1
WHEN count(ORDER_NUMBER) <= 5 AND a.CAT != 'E' THEN .05
WHEN count(ORDER_NUMBER) >= 6 AND count(ORDER_NUMBER) < 11 AND a.CAT != 'E' THEN .25
WHEN count(ORDER_NUMBER) >= 11 AND count(ORDER_NUMBER) < 21 AND a.CAT != 'E' THEN .5
WHEN count(ORDER_NUMBER) >= 21 AND a.CAT != 'E' THEN 1 END) as TSCORE
FROM trx.TRX_Import t3
JOIN ATTRIBUTES a
    ON a.ITEM = t3.ITEM
WHERE `TRANS_DATE` BETWEEN DATE_SUB(CURDATE(), INTERVAL 365 DAY) AND CURDATE()
AND t3.ITEM = 'WARI-BB150'
GROUP BY MONTH
) t2

Spits out:

Item         AvgTSCORE
WARI-BB150   0.55000