skip to main content skip to footer

Reduced Rank Classification

Author(s):
Dorans, Neil J.
Publication Year:
1985
Report Number:
RR-85-07
Source:
ETS Research Report
Document Type:
Report
Page Count:
38
Subject/Key Words:
Classification, Multivariate Analysis, Statistical Analysis

Abstract

Two reduced rank classification procedures, principal components classification and equal weights classification, are described and compared via a simulation study to the standard classification procedure to determine their feasibilities as alternative classification procedures. First, a justification for the development of these two reduced rank procedures is provided. Then, the two reduced rank rules are derived. The simulation design is described in detail. The simulation results demonstrate that the reduced rank procedures are preferable to the standard procedure under certain conditions (i.e., when they appropriately incorporate prior information about population structure into their classification rules). Suggestions for future research are offered.

Read More