View previous topic :: View next topic
|
Author |
Message |
Steve Ironmonger
New User
Joined: 19 Oct 2015 Posts: 15 Location: UK
|
|
|
|
Hi, first post, so please be gentle.
I've been working on some data mining and i'm stuck on one last issue.
Basically I'm creating a file with 3+ million records in it and I want to create a file with just a sample of those 3 million records. I've written a simple REXX to do this but as it's a REXX it's not that quick....i.e. it takes 15/20 mins to run and will use +50% of the cpu if the mainframe is quiet...
My rexx reads in 10000 records at a time, adds 1 to a counter and then each time the counter // 30 it then writes out the record to an array/stack
I then end up with a file 1/30th the size of the original and the records sampled from across the whole file
Can i do this in sort ?
I have had a look through the dfsort manuals and asked the Oracle (google) but it's like looking for a needle in a haystack !
thanks
Steve |
|
Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
|
|
|
|
You used the word "sample". If you look in the index, or search within the documentation for SAMPLE, you'll find it on OUTFIL, then you just need to arrange what you want. |
|
Back to top |
|
|
RahulG31
Active User
Joined: 20 Dec 2014 Posts: 446 Location: USA
|
|
|
|
Quote: |
I have had a look through the dfsort manuals |
If you have a look at the manual again, you'll find this (and more)
Quote: |
* PROCESS RECORDS 1, 6, 11, ...
OUTFIL FNAMES=OUT1,SAMPLE=5 |
|
|
Back to top |
|
|
Steve Ironmonger
New User
Joined: 19 Oct 2015 Posts: 15 Location: UK
|
|
|
|
Thanks guys, thats perfect....
It's knowing what to search for that's key to working these things out, maybe I need to simplify my searches ! |
|
Back to top |
|
|
|