Count whether a value exist from a range of keys - mysql

Question

Return a count when response_val=4 is found between response_key 95 to 98 corresponding to each uhid and group by month
|response_key | response_val | month | uhid |
|---------------------------------------------|
| 95 | 4 | 2019-09| 14569|
| 96 | 4 | 2019-09| 14569|
| 97 | 5 | 2019-09| 14569|
| 98 | 5 | 2019-09| 14569|
| 95 | 5 | 2019-09| 13256|
| 96 | 5 | 2019-09| 13256|
| 97 | 5 | 2019-09| 13256|
| 98 | 5 | 2019-09| 13256|
| 95 | 5 | 2019-09| 25689|
| 96 | 5 | 2019-09| 25689|
| 97 | 5 | 2019-09| 25689|
| 98 | 5 | 2019-09| 25689|
|---------------------------------------------|
I tried sql query as
SELECT month,
COUNT(CASE
WHEN `response_key` IN (95,96,97,98) and `response_val`='5'
THEN 1 ELSE NULL END) AS 'yes',
COUNT(CASE
WHEN `response_key` IN (81,82,83,84,85,86,87) and `response_val`='4'
THEN 1 ELSE NULL END) AS 'no'
FROM audit WHERE group by month
but it return the count when all the response_key have value 4.
I expect result as
| count | month |
-----------------
| 1 |2019-09|

Answer 1

Maybe the following will get you what you want:
SELECT `month`,
COUNT(CASE response_val WHEN 4 THEN 1 END) AS `yes`,
COUNT(CASE response_val WHEN 5 THEN 1 END) AS `no`
FROM audit
WHERE response_key between 95 AND 98
GROUP BY `month`
I simplified your case statements and move the response_key condition into the where-clause (the ELSE part in a CASE statement is only necessary if you want to return a value other than null).
You can find a working demo here: https://rextester.com/PCN24609
The resulting counts from this query are:
month yes no
2019-09 1 11
Well, you can swap the order of your output columns as you like, and if you are only interested in the response_val=4 counts do it like here:
SELECT COUNT(CASE response_val WHEN 4 THEN 1 END) AS `count`,
`month`
FROM audit
WHERE response_key between 95 AND 98
GROUP BY `month`
demo: https://rextester.com/JEV40175
result:
| count | month |
-------------------
| 1 | 2019-09 |
Edit:
"Return a count when response_val=4 is found between response_key 95
to 98 corresponding to each uhid and group by month"
This would imply that you simply group over uhid too, like shown here:
SELECT COUNT(CASE response_val WHEN 4 THEN 1 END) AS `count`,
`month`,uhid
FROM audit
WHERE response_key between 95 AND 98
GROUP BY `month`,uhid;
-- results:
count month uhid
0 2019-09 13256
2 2019-09 14569
1 2019-09 25689
Or, if you want all uhid-counts in one row per month:
SELECT `month`,
COUNT(CASE uhid WHEN 14569 THEN 1 END) AS `cnt_14569`,
COUNT(CASE uhid WHEN 13256 THEN 1 END) AS `cnt_13256`,
COUNT(CASE uhid WHEN 25689 THEN 1 END) AS `cnt_25689`
FROM audit
WHERE response_key between 95 AND 98 AND response_val=4
GROUP BY `month`,uhid
-- results:
month cnt_14569 cnt_13256 cnt_25689
2019-09 2 0 0
2019-09 0 0 1
demo: https://rextester.com/WXQ73318
2. Edit: Following your latest comment, the following might fulfil your requirements
("list the number of distinct users with response_val=4 for each month"):
SELECT COUNT(`count`) usrcnt, `month` FROM (
SELECT COUNT(CASE response_val WHEN 4 THEN 1 END) AS `count`,`month`,uhid
FROM audit WHERE response_key between 95 AND 98
GROUP BY `month`,uhid
) tbl WHERE `count`>0 GROUP BY `month`;
-- or, even shorter:
SELECT COUNT(DISTINCT CASE response_val WHEN 4 THEN uhid END) AS `count`,`month`
FROM audit WHERE response_key between 95 AND 98
GROUP BY `month`;
-- same result from both queries:
usrcnt month
2 2019-09
demo: https://rextester.com/UGN73913
The latest queries are based on these table values:
response_key response_val month uhid
95 4 2019-09 14569
96 4 2019-09 14569
97 5 2019-09 14569
98 5 2019-09 14569
95 5 2019-09 13256
96 5 2019-09 13256
97 5 2019-09 13256
98 5 2019-09 13256
95 5 2019-09 25689
96 4 2019-09 25689
97 5 2019-09 25689
98 5 2019-09 25689