SQL query : SELECT - sql

Question

CREATE TABLE WRITTEN_BY
( Re_Id CHAR(15) NOT NULL,
Pub_Number INT NOT NULL,
PRIMARY KEY(Re_Id, Pub_Number),
FOREIGN KEY(Re_Id) REFERENCES RESEARCHER(Re_Id),
FOREIGN KEY(Pub_Number) REFERENCES PUBLICATION(Pub_Number));
CREATE TABLE WORKING_ON
( Re_Id CHAR(15) NOT NULL,
Pro_Code CHAR(15) NOT NULL,
PRIMARY KEY(Re_Id, Pro_Code, Subpro_Code)
FOREIGN KEY(Re_Id) REFERENCES RESEARCHER(Re_Id));
Re_Id stands for ID of a researcher
Pub_Number stands for ID of a publication
Pro_Code stands for ID of a project
Written_by table stores information about a Publication's ID and it's author
Working_on table stores information about a Project's ID and who is working on it
Now, I have this query :
For each project, find the researcher who wrote the most number of publications .
This is what i've done so far :
SELECT Pro_Code,WORK.Re_Id
FROM WORKING_ON AS WORK , WRITTEN_BY AS WRITE
WHERE WORK.Re_Id = WRITE.Re.Id
so I got a table which contains personal ID and project's ID of a researcher who has at least 1 publication. But what's next ? How to solve this problem?

Answer 1

You haven't said which platform you're on but try this. It handles the case where there are ties as well.
select g.Pro_Code, g.Re_Id, g.numpublished
from
(
SELECT work.Pro_Code, WORK.Re_Id, count(WRITE.pub_number) as numpublished
FROM WORKING_ON WORK JOIN WRITTEN_BY AS WRITE ON WORK.Re_Id = WRITE.Re_Id
GROUP BY work.Pro_Code, WORK.Re_Id
) g
inner join
(
select Pro_code, max(numpublished) as maxpublished
from (
SELECT work.Pro_Code, WORK.Re_Id, count(WRITE.pub_number) numpublished
FROM WORKING_ON WORK JOIN WRITTEN_BY AS WRITE ON WORK.Re_Id = WRITE.Re_Id
GROUP BY work.Pro_Code, WORK.Re_Id
) g2
group by Pro_code
) m
on m.Pro_code = g.Pro_Code and m.maxpublished = g.numpublished
Some platforms will allow you to write it this way:
with g as (
SELECT work.Pro_Code, WORK.Re_Id, count(WRITE.pub_number) as numpublished
FROM WORKING_ON WORK JOIN WRITTEN_BY AS WRITE ON WORK.Re_Id = WRITE.Re_Id
GROUP BY work.Pro_Code, WORK.Re_Id
)
select g.Pro_Code, g.Re_Id, g.numpublished
from g
inner join
(
select Pro_code, max(numpublished) as maxpublished
from g
group by Pro_code
) m
on m.Pro_code = g.Pro_Code and m.maxpublished = g.numpublished

Answer 2

I think that you are looking for something like the following :
select
tm.pro_code as pro_code,
tm.re_id as re_id,
max(total) as max_pub
from (
select *
from (
select
wo.pro_code as pro_code
wr.re_id as re_id,
count(wr.pub_number) as total
from
written_by wr,
working_on wo
where
wr.re_id = wo.re_id
group by wr.re_id,wo.pro_code
)
) tm
group by pro_code

Answer 3

If you are using MS SQL, this should work:
With cte as (
select a.Re_Id, Pub_Number,Pro_Code, COUNT(distinct Pub_Number) as pubs
from WRITTEN_BY a
inner join WORKING_ON b
on a.Re_Id = b.Re_Id)
SELECT Re_Id,pubs from cte
HAVING pubs = MAX(pubs)
GROUP BY Re_Id