How to only select rows where one column is unique and the other is a certain value

Question

I am trying to figure out which ids only have one type of transaction.

I have tried joining and selecting distinct but I am doing it wrong

select transactions.type, id 
from datasource
group by id, transactions.type

This gives me a table with two transaction types: either dep or withdraw. Most IDs have two rows, one being dep and the other withdraw. I want to select only the ids that have only the withdraw transaction type

Answer 1

use where condition and not exists

select transactions.type, id from datasource t1
where type='withdraw'
and not exists( select 1 from datasource t2 where t1.id=t2.id
               and type='dep')

Answer 2

You can use group by with having count(*)=1 clause

select id 
  from datasource 
 where transactions_type in     
    (
     select transactions_type, id 
       from datasource
      group by transactions_type
     having count(*)=1 )

Since "You are trying to figure out which ids only have one type of transaction"

If you specifically look for transactions_type = 'withdraw' , then add and transactions_type = 'withdraw' to the end of the above Select Statement.

PS I suppose there's a type for transactions.type , and that should be transactions_type , isn't that?

Answer 3

I think aggregation does what you want. However, I am confused what your data structure is. A join is needed somewhere:

select ds.id, min(t.type)
from datasource ds join
     transactions t
     on ds.? = t.?   -- what is the join key?
group by ds.id
having min(t.type) = max(t.type);

If transactions.type is actually a column in datasource , then:

select ds.id, min("transactions.type")
from datasource ds 
group by ds.id
having min("transactions.type") = max("transactions.type");

Answer 4

Similar to other answers, but simpler:

select id, count(distinct transactions.type)
from datasource
group by id
having count(distinct transactions.type) = 1

It is unclear, however, what transactions.type is. It doesn't appear to be valid as a column name. Or did you mean to write transactions_type ?