Question

I’ve successfully merged 3 tables (using ID’s), but when I try to add 4th table – output completely messes up (values become unrealistic/wrong) so I think that I has something to do with the fact, that this 4th table has ungrouped ID’s, so I need to group them before Joining this new table. Right now the query is following:

SELECT name, SUM(money) AS MONEY
FROM transactions
JOIN results ON transactions.id = results.id
JOIN more ON results.per_id = more.per_id
GROUP BY name
HAVING SUM(money)>500

and when I join new table:

SELECT name, SUM(money) AS MONEY, SUM(data_from_NT1), SUM(data_from_NT2)
FROM transactions
JOIN results ON transactions.id = results.id
JOIN more ON results.per_id = more.per_id

JOIN newtable ON results.per_id = newtable.per_id

GROUP BY name
HAVING SUM(money)>500

is it possible to execute a command GROUP BY per_id:

(JOIN newtable ON results.per_id = newtable.per_id GROUP BY per_id)

before adding this new table into the main table? The line from above doesn’t work.

Answer 1

Yes, it is possible, but you have to write it as a SELECT. This is a first version of a possible answer:

SELECT name,
       SUM(money)             AS money,
       SUM(nt2.data_from_NT1) AS data_from_NT1,
       SUM(nt2.data_from_NT2) AS data_from_NT2
  FROM transactions AS t
  JOIN results      AS r ON t.id = r.id
  JOIN more         AS m ON r.per_id = m.per_id
  JOIN (SELECT per_id, SUM(data_from_NT1) AS data_from_NT1, SUM(data_from_NT2) AS data_from_NT2
          FROM newtable GROUP BY per_id
       ) AS nt2 ON results.per_id = nt2.per_id
 GROUP BY name
HAVING SUM(money) > 500;

It isn’t completely clear whether you need the ‘SUM of SUMs’ in the main select-list; you probably don’t. It also isn’t clear where the name or money columns come from; I’d normally prefix them with the appropriate table alias, too. With those caveats, this might be more nearly what you’re after:

SELECT name, SUM(money) AS MONEY, nt2.data_from_NT1, nt2.data_from_NT2
  FROM transactions AS t
  JOIN results      AS r ON t.id = r.id
  JOIN more         AS m ON r.per_id = m.per_id
  JOIN (SELECT per_id, SUM(data_from_NT1) AS data_from_NT1, SUM(data_from_NT2) AS data_from_NT2
          FROM newtable GROUP BY per_id
       ) AS nt2 ON results.per_id = nt2.per_id
 GROUP BY name, nt2.data_from_NT1, nt2.data_from_NT2
HAVING SUM(money) > 500;

There are undoubtedly other ways to write this. The grouping in the main query might be better put into a parallel sub-query, leaving simple direct joins in the main query. But we don’t have the information to actually do that for you.