SQL to select common as well as remaining data using JOIN in MySQL

Question

I have three tables: wi_district, wi_group, and wi_training. I need to count groups and trainings based on districts. For this purpose I have used following SQL;

• SQL1

SELECT wi_district.dst_name, COUNT(grp_id) AS group_count, MAX(grp_created_date) as grp_created_date 
FROM wi_group 
INNER JOIN wi_district ON wi_district.dst_id=wi_group.grp_dst_id AND wi_group.grp_deleted=0 AND wi_group.grp_type IN (3)
GROUP BY wi_district.dst_name

The query counts group for each district. Likewise,

• SQL2

SELECT wi_district.dst_name, COUNT(trn_id) AS training_count, MAX(trn_created_date) as trn_created_date 
FROM wi_training 
INNER JOIN wi_district ON wi_district.dst_id=wi_training.dst_id AND wi_training.trn_deleted=0 AND wi_training.trn_beneficiary_type IN (-1,2,8,9,10)
GROUP BY wi_district.dst_name

The query counts training for each district. Now I need to combine all the results obtained from SQL1 and SQL2 and get the result in the form of

dst_name || group_count || grp_created_date || training_count || trn_created_date

The problem is whenever I used SQL1 LEFT JOIN SQL2 then it displayed the result respective to SQL1 where the result of SQL2 can't be obtained and vice-versa. Please help me sort out with this problem in MySQL

Answer 1

I think you can join the filtered tables and then group by the district name. Like so:

SELECT dist.dst_name AS dst_name,
COUNT(grp.grp_id) AS group_count, MAX(grp.grp_created_date) AS grp_created_date,
COUNT(trn.trn_id) AS training_count, MAX(trn.trn_created_date) AS trn_created_date 
FROM wi_district AS dist
LEFT JOIN (
  SELECT dst_id, trn_id, trn_created_date
  FROM wi_training
  WHERE trn_deleted=0
    AND trn_beneficiary_type IN (-1,2,8,9,10)
) AS trn ON trn.dst_id=dist.dst_id
LEFT JOIN (
  SELECT grp_dst_id, grp_id, grp_created_date
  FROM wi_group
  WHERE grp_deleted=0
    AND grp_type IN (3)
) AS grp ON grp.grp_dst_id = dist.dst_id
GROUP BY dist.dst_name

Answer 2

Why use names like dst_name? It is better to write out the full name, so you will also understand what it means after a few months.

Anyway, this query should do the trick

select d.dst_name as district
,      count(distinct g.grp_id) as group_count
,      max(grp_created_date) as group_created_date 
,      count(distinct trn_id) as training_count
,      max(trn_created_date) as trn_created_date 
from   wi_district d
left join wi_group g on  d.dst_id = g.grp_dst_id
                     and g.grp_deleted = 0 
                     and g.grp_type in (3)
left join wi_training t on  d.dst_id = t.dst_id
                        and t.trn_deleted = 0
                        and t.trn_beneficiary_type IN (-1,2,8,9,10)
group by d.dst_name

Answer 3

You need to DRIVE it from the grouping table. Like so.

SELECT  D.dst_name,  COUNT(grp_id) AS group_count, MAX(grp_created_date) as grp_created_date, COUNT(trn_id) AS training_count, MAX(trn_created_date) as trn_created_date 
FROM    wi_district D
LEFT JOIN wi_group G ON D.dst_id = G.grp_dst_id AND G.grp_deleted=0 AND G.grp_type IN (3)
LEFT JOIN wi_training T ON  D.dst_id = T.dst_id AND T.trn_deleted=0 AND T.trn_beneficiary_type IN (-1,2,8,9,10)
GROUP BY wi_district.dst_name

And if you only want rows if they exist on either table, add a clause:

WHERE NOT G.grp_dst_id IS NULL OR NOT D.dst_id IS NULL