topic Re: Clustering in SAS Programming

Clustering

Himanshu007 — Sat, 03 Jun 2017 07:55:23 GMT

I have the following data

PS Lat Long SLOT

abc 75.8233 26.2454 1
abc 75.4785 26.4589 1

abc 75.5495 26.4679 1
abc 75.4875 26.8997 2
abc 75.5649 26.7268 2
abc 75.5666 26.8647 2
xyz 75.4875 26.3842 1
xyz 75.4937 26.5877 1
xyz 75.8473 26.3971 1
xyz 75.7861 26.6884 2

Now I want a new column which shows the distances in the following way:

All the points with PS=abc and Slot=1 should have their distances calculated from the first point of this group(i.e., 75.8233 26.2454), similarly all the points with PS=abc and Slot=2 should have their distances calculated from the first point of this group(i.e., 75.4875 26.8997) and so on.

So please suggest me a solution for this.

Re: Clustering

PeterClemmensen — Sat, 03 Jun 2017 14:16:24 GMT

Check out the GEODIST function

http://support.sas.com/documentation/cdl/en/lefunctionsref/63354/HTML/default/viewer.htm#n1korpfg2e18lon1nwpow9qijdxe.htm

Re: Clustering

Himanshu007 — Mon, 05 Jun 2017 07:25:10 GMT

But this function is only there for finding the distances my concern is how to fix one point from each group(group means same PS and same Slot) as a fixed reference point from which the other distances are to be found

Re: Clustering

ballardw — Mon, 05 Jun 2017 19:41:25 GMT

Assuming your data are sorted by PS and Slot as implied by the example:

data want;
   set have;
   by ps slot;
   retain firstLat Firstlong;
   if first.slot then do;
      firstlat=lat;
      firstlong=long;
   end;
   Dist = geodist(firstlat,firstlong,lat,long);
   drop firstlat firstlong;
run;