? MS ACCESS 2010 sample to calculate median quickly?

I have a data base with several payments for each participant.
Each payment has participant number and date-of-payment.

I'd like to easily and QUICKLY calculate the Median number of days between payments for each participant.

I found a method using a sub-query, but, it ran for over 12 hours on only about 30,000 rows.
I'd like to be able to run it on a half-million rows in less than an hour.

help?
LVL 1
philkryderAsked:
Who is Participating?
 
Patrick MatthewsCommented:
Have you had a chance to review the approach described in my article above?  (Thanks for referring it, Dale!)
0
 
Dale FyeCommented:
What did your query look like?

To start with, you have to get the payment date and the previous payment date,  I generally do that something like:

SELECT P1.ParticipantID, P1.PaymentDate, Max(P2.PaymentDate) as PrevPayment
FROM yourTable as P1
LEFT JOIN yourTable as P2
ON P1.ParticipantID = P2.ParticipantID
AND P1.PaymentDate >P2.PaymentDate
GROUP By P1.ParticipantID, P1.PaymentDate

This should run relatively quickly, but is not editable in the query grid, only in the SQL view.  Once you have that working, you can either save it, or make it a subquery to get the difference in dates between the PaymentDate and PrevPayment.  You might want to make that a temp table or even make the above into a temp table, to speed up succeeding logic.

What logic are you using to compute the median for each participant, once you know the number of days between payments?
0
 
Dale FyeCommented:
BTW, you might want to check out this article by MatthewsPatrick that discusses Median, Mode, Skewness, and Kurtosis
0
Ultimate Tool Kit for Technology Solution Provider

Broken down into practical pointers and step-by-step instructions, the IT Service Excellence Tool Kit delivers expert advice for technology solution providers. Get your free copy now.

 
philkryderAuthor Commented:
yes - I used the DMEDIAN on an INTERVAL value that I obtained using the code Dale showed for finding the interval between the two dates.

After finding the INTERVAL it would find the DMEDIAN.

I saved intermediate tables after each step.

I was running this on an I3 intel and it pegged 100% of one of the CPU engines for about 12 hours to do my 32k rows.


My typical grouping on Participant will have from 7 to 20 intervals.


Have you run this on similar data quantities?

I'll post the code when I get back to work after the holiday tomorrow.
Phil
0
 
Patrick MatthewsCommented:
A sample file would be very helpful
0
 
philkryderAuthor Commented:
I'd like to close this.
On closer examination, I found the major delay was in calculating the interval between two timestamps.

I subsequently calculated the interval by sorting in order by the key, and the timestamp, then assigning a sequential ID, then doing a self-join on the next row. This seemed much faster than the sub-query to find the interval.

Matthew -
regarding the median -
would you consider adding a couple of global variables such that if the CRITERIA was the same between calls to DMEDIAN, then the DMEDIAN returned would be the same as on the prior call.
Not as elegant,
But, It seems it would avoid creating and running the subquery much of the time  ...
something like:
' =================
Global DMedianSave As Long
Global CriteriaSave As String
...

    If Criteria <> "" Then
   
        If CriteriaSave = Criteria Then
            DMedian = DMedianSave
            GoTo Cleanup
        End If
    End If
    CriteriaSave = Criteria
...
 '        calculate DMEDIAN as before
...
   End If        
    DMedianSave = DMedian
    GoTo Cleanup
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.