How to get the reverse result to a sql query

Question

I am using the following sql query to get the name of a branch that has submitted a report.

SELECT DISTINCT branch.b_name AS branch from branch, report_activity2

where branch.branch_id=report_activity2.branch_id 

and week =16
and year =2022

What I would also like to do is get the names of branches that have not submitted a report for the same week and year. I have tried the following query:

SELECT branch.b_name FROM branch

WHERE branch_id not in

(SELECT DISTINCT branch.b_name AS branch from branch, report_activity2

where branch.branch_id=report_activity2.branch_id 

and week =16
and year =2022)

But the second query result lists all the branches and does not exclude the branch from the first query. I am not sure if I am doing this completely wrong or I have just missed something simple. Any help to nudge me in the right direction would be appreciated.

Answer 1

In the first one you don't really need distinct and also you shouldn't use the old style join anyway:

SELECT b_name AS branch from branch
where exists (select * from report_activity2
where branch.branch_id=report_activity2.branch_id
and report_activity2.week =16
and report_activity2.year =2022);

In the second one you should compare branch_id to branch_ or b_name to b_name (provided it is unique) but you were checking branch_id in (... b_bame ...). Anyway it also doesn't need a join:

SELECT b_name AS branch from branch
where NOT exists (select * from report_activity2
where branch.branch_id=report_activity2.branch_id
and report_activity2.week =16
and report_activity2.year =2022);

EDIT: Some more explanation. With EXISTS (...), per row of branch, we are checking if in report_activity2 there is a row where branch_id matches "current" branch.branch_id and week = 16 and year 2022. If a match is found then it acts as a true on filtering branch. The good part is, there might be 1 or 1000 matching rows, it wouldn't matter and unlike a join it would just stop finding the first match thus more performant than doing a join (in which case you also would need to have DISTINCT, if there were 1000 rows matching a particular branch.branch_id, you first get 1000 rows and then DISTINCT select 1 one them). Somewhat visual explanation:

Branch:

Branch_id, b_name
1, 'b1'
2, 'b2'

Report_Activity2:

Branch_Id, Week, Year, ...
1,  1, 2022
1, 16, 2022
1, 16, 2022
1, 16, 2022
2,  1, 2022

With join it looks like (3 matches for 1):

1, 'b1', 1, 16, 2022
1, 'b1', 1, 16, 2022
1, 'b1', 1, 16, 2022

and then you use DISTINCT on b_name to get 'b1'.

With exists:

On

  1, 'b1' 

a match is found thus it immediately returns a true on first

1, 16, 2022 

without checking other rows in report_activity2. Resulting set is already:

1, 'b1'

Thus you don't need a DISTINCT hack.

Answer 2

can you try this mysql command ?

SELECT DISTINCT branch.b_name AS branch from branch, report_activity2

where branch.branch_id != report_activity2.branch_id 

and week =16
and year =2022 ;