Potential Approaches / Solutions

The Outlier procedure
Computes the ratio of the highest dollar transaction with the amount the next to the highest dollar transaction. High ratios indicate the potential for miskeying or overstatement. For example, invoices from one particular vendor may tend to be bunched around a central figure, but one invoice was miskeyed, for example $1,995 instead of $19.95.

Syntax

proc outlier data=DATAFILE factor=FACTOR;
var VARIABLENAME;
out = OUTPUTRESULT;
by BYVAR;

Parameters Used

There are five parameters for the RV procedure:

DATAFILE - the datafile to be analyzed
FACTOR - the cutoff amount to select records (must be 1.0 or larger)
VARIABLENAME - the numeric variable to be analyzed
OUTPUTRESULT - name of the file to store the results of the analysis
BYVAR - the field used to identify a control break, e.g. supplier number, employee number etc.

Example Script

*;
* RV.ezs;
* determine if any area in the census table had a significant;
* difference between the highest and next highest;
* number of rental households (v0031990) in the 1990 census;
* cutoff point is 200%;
libname test '{%libout}';
libname rv '{%libin}';
proc RV data=rv.censuspart1 factor=2;
out = test.rv1;
var v0031990;
by areaname;
run;

View Example Output

 

The Proc CR procedure is part of EZ-R Stats for Windows - Standard Edition. View an overview of EZ-R Stats for Windows. Download EZ-R Stats for Windows and other products.

Analysts interested in Proc CR may also be interested in Invoice Analysis and Proc Domain.