Question

I am very new to using SQL and I haven’t been able to find any previous posts that seem to answer my question. I need a statement that will give me the latest approved submission for every customer in the database and all rows/columns of data associated with that submission. Each submission is assigned a unique id number and is in a column called ASMT_ID. The unique customer id number is a column called PRTPT_ID.

So, for each PRTPT_ID in the database T_FINANCIAL_ITEM, I want to find the MAX ASMT_ID for each customer and display all columns and rows of data associated with each PRTPT_ID, but only if the STTS_NAME is equal to “Approved”.

In my simple terms what I’m trying to do:

select * from T_FINANCIAL_ITEM
For each PRTPT_ID SELECT MAX ASMT_ID
WHERE STTS_NAME = 'Approved'
GROUP by PRTPT_ID

*UPDATE*
Rather than using the asmt_id, I will need to analyze two other fields in the same table FYE_DT and SUBM_TYPE.

The criteria is:

1. The most recent FYE_DT with a stts_name ‘Approved’ (i.e. 2011 before 2010).
2. In the most recent fye_dt, the audited subm_type
3. In the most recent fye_dt, if their is no audited subm_type, the unaudited subm_type
4. It will pull all rows of data associated with the submission meeting the criteria

For example:

PRTPT_ID    ASMT_ID FYE_DT                       SUBM_TYPE_NAME  
8493000000  18016   30-JUN-09 12.00.00.000000000 AM Unaudited/A-133  
8493000000  19574   30-JUN-09 12.00.00.000000000 AM Audited/A-133  
8493000000  28039   30-JUN-10 12.00.00.000000000 AM Unaudited/A-133  
8493000000  33620   30-JUN-10 12.00.00.000000000 AM Audited/A-133  
9481000000  38049   31-DEC-10 12.00.00.000000000 AM Unaudited/Non-A133  
9481000000  58020   31-DEC-09 12.00.00.000000000 AM Audited/Non-A-133  
9481000000  48139   31-DEC-11 12.00.00.000000000 AM Unaudited/Non-A-133  

Result:

PRTPT_ID    ASMT_ID FYE_DT                       SUBM_TYPE_NAME  
8493000000  33620   30-JUN-10 12.00.00.000000000 AM Audited/A-133  
9481000000  48139   31-DEC-11 12.00.00.000000000 AM Unaudited/Non-A-133  

Note: There can be multiple rows of data associated with these results. I need to have all of the rows of data.

Answer 1

One relatively standard approach would be to use analytic functions. If I understand your update, it sounds like you want the ORDER BY clause to be something like

select *
  from (select t.*,
               rank() over (partition by prtpt_id
                                      order by fye_dt desc,
                                               (case when subm_type_name like 'Audited%'
                                                     then 2
                                                     when subm_type_name like 'Unaudited%'
                                                     then 1
                                                     else 0
                                                  end) desc) rnk
          from t_financial_item t
         where stts_name = 'Approved')
 where rnk = 1