topic Re: using dbscan, cluster, to indentify outliers in Statistical Procedures

using dbscan, cluster, to indentify outliers

geneshackman — Tue, 09 Jun 2020 16:08:19 GMT

Question about dbscan. I am -just- learning about it, so I'm not even sure I can use it.

I have indicators by county, for example, total mortality rates. Supppose I have a list of counties and their values, I'll just list some of them here:

23 640

12 642

44 775

56 897

57 4,484

Looks like county 57 is outside the cluster of other counties.

Is this the sort of thing for which I can use dbscan or cluster analysis? Is there a 'gentle' introduction to dbscan or cluster analysis? I got this

https://support.sas.com/documentation/cdl/en/inmsref/67306/HTML/default/viewer.htm#p08w4hxeqsgklrn14aqs76gs01te.htm

Anything more introductory?

Re: using dbscan, cluster, to indentify outliers

geneshackman — Tue, 09 Jun 2020 16:10:49 GMT

I forgot to say that the first column is county, the second column is rate.

Re: using dbscan, cluster, to indentify outliers

Ksharp — Wed, 10 Jun 2020 12:26:57 GMT

Could try Primary Component Analysis .

proc pls

proc pca

@Rick_SAS wrote blogs about how to identity outliers .

Re: using dbscan, cluster, to indentify outliers

geneshackman — Thu, 11 Jun 2020 16:50:32 GMT

Thanks, I will look at primary components analysis.

Gene