An OR Tool for Partitioning Datasets into Uniform Segments

PDF

Authors
  1. Emond, E.J.
Corporate Authors
Defence R&D Canada - Centre for Operational Research and Analysis, Ottawa ON (CAN)
Abstract
This report describes an operational research tool for dividing a set of scalar values into categories or clusters without having to specify the number of clusters in advance. The method has a theoretical underpinning based on a statistical test for outliers of a uniform distribution derived by D.A. Darling in 1952. The problem arose in an operational research study for the Canadian Department of National Defence in which it was necessary to segment a set of 82 options into categories based on a numerical score. The theoretical background is described and examples are given to illustrate the methodology.g8070
Keywords
clustering;outliers;uniform distribution;ordered categories
Report Number
DRDC-CORA-TN-2009-052 — Technical Note
Date of publication
01 Nov 2009
Number of Pages
22
DSTKIM No
CA046267
CANDIS No
807029
Format(s):
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: