CODEWITHSUNDEEP
sqloracle-databasedate
bawpie
bawpie

Reputation: 477

Returning nearest date to date in a different table

I have two tables, one called Events and one called Documents. Each table has two date columns, start and end (so Event Start, Event End, Doc Start, Doc End). Both tables are linked by a field called Customer_ID. I'm interested in returning the closest relevant event start (and its ID) date to the Doc start date based on the customer ID, so for example the data in the tables might look like:

Customer_ID DOC_ID      DOC_Start   DOC_END
A           12          22/01/2011  23/01/2011
A           13          01/12/2011  05/12/2011
C           22          13/03/2011  20/03/2011


Customer_ID Event_ID    Event_Start Event_END
A           J1          01/01/2011  23/01/2011
A           J2          04/12/2011  05/12/2011
C           J44         15/03/2011  20/03/2011

I'd want the final result to show this:

Customer_ID DOC_ID  DOC_Start   DOC_END     Event_ID
A           12      22/01/2011  23/01/2011  J1
A           13      01/12/2011  05/12/2011  J2
C           22      13/03/2011  20/03/2011  J44

I've tried googling for a solution, and tried one or two suggestions but none of the examples I can find seem to be for comparing dates across tables, or even for Oracle. Also, I have fairly limited knowledge when it comes to SQL, but most that I have picked up has been from here, so thanks already for the assistance I've already received.

Edit: A further condition is that I'd like to return correlating event/doc types only. So I only want to bring back DOC_ID's where the DOC_TYPE is 'Enquiry' or 'Info', and the same goes for the Event_Type's.

Edit: Thanks Vincent, I put the where in as suggested (think I'd tried it everywhere but there!) and I'm now getting the results required.

I'll post the working solution for future reference if anyone is interested:

SELECT
*

FROM (SELECT 
O_ASSESSMENTS.ASM_SUBJECT_ID as "ID", 
O_ASSESSMENTS.ASM_ID as "Assessment ID", 
O_ASSESSMENTS.ASM_START_DATE as "Assessment Start",  
O_ASSESSMENTS.ASM_END_DATE as "Assessment End", 
O_SERVICE_EVENTS.SEV_ID as "Event ID", 
O_SERVICE_EVENTS.SEV_ACTUAL_DATE as "Event Start", 
O_SERVICE_EVENTS.SEV_OUTCOME_DATE as "Event End",
ROUND(O_ASSESSMENTS.ASM_START_DATE -O_SERVICE_EVENTS.SEV_ACTUAL_DATE,0) as "Likely",
  row_number() over(PARTITION BY  O_ASSESSMENTS.ASM_ID                                    
ORDER BY abs(O_ASSESSMENTS.ASM_START_DATE -  O_SERVICE_EVENTS.SEV_ACTUAL_DATE)) rn            
FROM 
O_ASSESSMENTS 
JOIN  O_SERVICE_EVENTS ON O_ASSESSMENTS.ASM_SUBJECT_ID = O_SERVICE_EVENTS.SEV_SUBJECT_ID
Where O_SERVICE_EVENTS.SEV_CODE IN ('t','t1') AND O_ASSESSMENTS.ASM_QSA_ID IN ('test','test1')  )  WHERE rn = 1

Upvotes: 2

Views: 7950

Answers (2)

Vincent Malgrat
Vincent Malgrat

Reputation: 67722

You can use analytics:

SELECT *
  FROM (SELECT d.customer_id, d.doc_id, d.doc_start, d.doc_end, e.event_id,
                row_number() over(PARTITION BY d.doc_id 
                                  ORDER BY abs(d.doc_start - e.event_start)) rn
           FROM doc d
           JOIN event e ON d.customer_id = e.customer_id
        /*WHERE CLAUSE*/)
 WHERE rn = 1

Upvotes: 8

Raihan
Raihan

Reputation: 10405

Join Events and Documents and only select joined rows with closest DOC_Start and Event_Start

select D1.Customer_ID, D1.DOC_ID, D1.DOC_Start, D1.DOC_END, E1.Event_ID 
from Events E1, Documents D1
where E1.Customer_ID = D1.Customer_ID
and (D1.Customer_ID, D1.DOC_ID, E1.Event_ID) in
(
     select D2.Customer_ID, D2.DOC_ID, E2.Event_ID 
     from Events E2, Documents D2
     where E2.Customer_ID = D2.Customer_ID
     group by D2.Customer_ID, D2.DOC_ID, E2.Event_ID
     having abs(D2.DOC_Start - E2.Event_Start) = min(abs(D2.DOC_Start - E2.Event_Start))
);

Upvotes: 0

Related Questions