* --------------------------------------------------------------------
	File: cluster.crime

	Purpose: cluster analysis of the crime data set
* -------------------------------------------------------------------;

OPTIONS NODATE NOCENTER PAGENO=1 LINESIZE=120;
* - NOTE: You must assign the LIBNAME p7291 to the directory/folder
	containing the data set crime;
*LIBNAME p7291 '';
DATA temp;
	SET p7291.crime;
RUN;

TITLE Cluster Analysis of the Crime data set;
TITLE2 'Average Linkage Method';
PROC CLUSTER DATA=temp METHOD=average STD
	PSEUDO RSQUARE OUTTREE=treedata;
  VAR murder--auto;
  ID state;
RUN;

*- the TREE procedure produces a dendrogram;
PROC TREE DATA=treedata HORIZONTAL;
  ID state;
RUN;

* - You should try different methods (see the sas documentation
	for the METHOD option for PROC CLUSTER) and compare the
	results with one another;

TITLE2 'Nearest Neighbor Density';
PROC CLUSTER DATA=temp METHOD=DENSITY K=3 STD OUTTREE=treedata;
  VAR murder--auto;
  ID state;
RUN;

*- the TREE procedure produces a dendrogram;
PROC TREE DATA=treedata HORIZONTAL;
  ID state;
RUN;