SQL Subquery all data from same table - sql

Question

I have a table where I would like to have one field (account) always shown then subqueries for counts or sums with criteria.
Example:
select ndhist_acct_nbr,
(select count(ndhist_acct_nbr) from dbo.nd_history where ndhist_type = '30'
and ndhist_rsn = '0' and ndhist_trcd = 'NF*' and ndhist_ref_type = '0' and ndhist_dt >= '03/01/2013') as NSF_TOTAL,
(select sum(ndhist_amt) from dbo.nd_history where ndhist_type = '30'
and ndhist_rsn = '98' and ndhist_trcd = 'TW0' and ndhist_ref_type = '11' and ndhist_dt >= '03/01/2013') as SIG_SPEND,
(select count(ndhist_acct_nbr) from dbo.nd_history where ndhist_type = '30'
and ndhist_rsn = '23' and ndhist_trcd = 'TW0' and ndhist_ref_type = '11' and ndhist_dt >= '03/01/2013') as PIN_TRANS,
(select count(ndhist_acct_nbr) from dbo.nd_history where ndhist_type = '30'
and ndhist_rsn = '21' and ndhist_trcd = 'SC*' and ndhist_ref_type = '0' and ndhist_dt >= '03/01/2013') as FOREIGN_AMT_FEE
from dbo.nd_history
group by ndhist_acct_nbr
The problem is the results - all of the account numbers show up but the counts/sum fields all repeat the data. Any help would be awesome!

Answer 1

Your subqueries are standalone - they don't depend on ndhist_acct_nbr field in any way, so the result is always the same.
Besides, this technique (using this many subqueries for every row of output) is a bad idea.
You should simplify the query, and instead of count distinct and subqueries, make sum(case when ... clauses.

Answer 2

Try:
select ndhist_acct_nbr,
count(case when ndhist_type = '30' and ndhist_rsn = '0' and ndhist_trcd = 'NF*' and ndhist_ref_type = '0' and ndhist_dt >= '03/01/2013'
then ndhist_acct_nbr end) as NSF_TOTAL,
sum(case when ndhist_type = '30' and ndhist_rsn = '98' and ndhist_trcd = 'TW0' and ndhist_ref_type = '11' and ndhist_dt >= '03/01/2013'
then ndhist_amt end) as SIG_SPEND,
count(case when ndhist_type = '30' and ndhist_rsn = '23' and ndhist_trcd = 'TW0' and ndhist_ref_type = '11' and ndhist_dt >= '03/01/2013'
then ndhist_acct_nbr end) as PIN_TRANS,
count(case when ndhist_type = '30' and ndhist_rsn = '21' and ndhist_trcd = 'SC*' and ndhist_ref_type = '0' and ndhist_dt >= '03/01/2013'
then ndhist_acct_nbr end) as FOREIGN_AMT_FEE
from dbo.nd_history
group by ndhist_acct_nbr
You can use derived columns from the results by putting the whole query inside an inline query and then selecting from it - like so:
select sq.*,
NSF_TOTAL*5 + SIG_SPEND*0.10 + PIN_TRANS*0.05 + FOREIGN_ATM_FEE as TOTAL_INCOME
from
(select ndhist_acct_nbr,
count(case when ndhist_type = '30' and ndhist_rsn = '0' and ndhist_trcd = 'NF*' and ndhist_ref_type = '0' and ndhist_dt >= '03/01/2013'
then ndhist_acct_nbr end) as NSF_TOTAL,
sum(case when ndhist_type = '30' and ndhist_rsn = '98' and ndhist_trcd = 'TW0' and ndhist_ref_type = '11' and ndhist_dt >= '03/01/2013'
then ndhist_amt end) as SIG_SPEND,
count(case when ndhist_type = '30' and ndhist_rsn = '23' and ndhist_trcd = 'TW0' and ndhist_ref_type = '11' and ndhist_dt >= '03/01/2013'
then ndhist_acct_nbr end) as PIN_TRANS,
count(case when ndhist_type = '30' and ndhist_rsn = '21' and ndhist_trcd = 'SC*' and ndhist_ref_type = '0' and ndhist_dt >= '03/01/2013'
then ndhist_acct_nbr end) as FOREIGN_AMT_FEE
from dbo.nd_history
group by ndhist_acct_nbr) sq
This could be done more elegantly via a CTE if you are using a RDBMS (such as Oracle, PostgreSQL or SQLServer) that supports CTEs.