Hamid, Hashibah and Mahat, Nor Idayu (2010) Some investigations on the LDA when handling large number of variable. In: Proceedings of Applied Mathematics International Conference 2010 (AMIC 2010) & The Sixth EASIAM Conference ), June 22th - 24th, 2010, PNB Darby Park, MALAYSIA. (Unpublished)

PDF Restricted to Registered users only Download (177kB)

Abstract

Linear discriminant analysis (LDA) has been used widely in many classification problems.This paper discusses the performances of LDA when the classification problems face with large number of variables.Two common strategies for constructing LDA are investigated: (i) some selected variables are used and (ii) all variables are combined systematically, in such a way that the performance of LDA is optimised.These strategies are studied on some example data sets through the leave-one-out procedure.The results indicate that, the performance of LDA with the combination of variables is the best based on the leave-one-out error rate.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	leave-one-out error rate, linear discriminant analysis, principal component analysis, variable selection approaches.
Subjects:	Q Science > QA Mathematics
Divisions:	College of Arts and Sciences
Depositing User:	Mrs. Norazmilah Yaakub
Date Deposited:	29 Feb 2012 09:41
Last Modified:	29 Feb 2012 09:41
URI:	https://repo.uum.edu.my/id/eprint/5193

Actions (login required)

View Item