Nisa, Khoirin (2009) PEMBANDINGAN BEBERAPA PENDUGA TINGKAT KESALAHAN KLASIFIKASI PADA ANALISIS DISKRIMINAN KUADRATIK. PROSIDING Seminar Sehari Hasil – Hasil Penelitian & Pengabdian Kepada Masyarakat Universitas Lampung. ISSN ISBN 978-979-8510-07-6

Nisa 2009.pdf

Download (553kB) | Preview


The major objective of discriminant analysis is to classify multivariate data into different population on the basis of a training sampel for which the source populations are known. Since the primary goal of discriminant analysis is classifying data, it is important to know the probability of misclassification (which is also called: classification error rate) of the classification rule we use. In this paper we compared four methods for estimating the classification error rate through Monte Carlo simulation, the methods are the Resubstituion method, the Jackknife method, U estimator and U estimator. We set the simulation using 1000 random samples with size: n = 20, 40 and 60. The comparison of the predictions of error rate was done using the MSE (mean square error) resulted from all methods. Te result showed that the Jackknife method always performs better than the other three methods.

Item Type: Article
Subjects: H Social Sciences > HA Statistics
Divisions: Fakultas Matematika dan Ilmu Pengetahuan Alam (FMIPA) > Prodi Matematika
Depositing User: DR. KHOIRIN NISA
Date Deposited: 27 Mar 2018 04:43
Last Modified: 27 Mar 2018 04:43

Actions (login required)

View Item View Item