Feature selection based on statistical analysis

Nooritawati Md Tahir, Aini Hussain, Salina Abdul Samad, Hafizah Husain, Mohd. Yusoff Jamaluddin

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

In most pattern recognition (PR) system, selecting the best feature vectors is an important task. Feature vectors serve as a reduced representation of the original data that facilitate us to evade the curse of dimensionality in a PR task. In this work, we deem further endeavor in selecting the best feature vectors for the PR task that is to determine the best eigenfeatures of four main human postures based on the rules of thumb of Principal Component Analysis namely the KG-rule, Cumulative Variance and the Scree Test. Accordingly, all three rules of thumb suggest in retaining only 9% of the total eigenvectors or also known as 'eigenpostures'. Next, these eigenpostures are statistically analyzed prior to classification. Thus, the most relevant component of the selected eigenpostures can be ascertained. The statistical significance of the eigenpostures is determined using ANOVA. Further, a Multiple Comparison Procedure (MCP) and homogeneous subsets tests are performed to determine the number of optimized eigenpostures for classification. These optimized eigenpostures will feat as inputs to the Artificial Neural Network (ANN) classifier. The statistical analysis has enabled us to perform effectively the selection of eigenpostures for classification of human postures.

Original languageEnglish
Pages (from-to)426-433
Number of pages8
JournalEuropean Journal of Scientific Research
Volume14
Issue number3
Publication statusPublished - 2006

Fingerprint

pattern recognition
Feature Vector
Feature Selection
Pattern Recognition
Statistical Analysis
Feature extraction
Statistical methods
statistical analysis
posture
Posture
Pattern recognition
Automated Pattern Recognition
Multiple Comparison Procedures
Pattern recognition systems
Curse of Dimensionality
Statistical Significance
Analysis of variance (ANOVA)
Principal Component Analysis
Eigenvalues and eigenfunctions
Principal component analysis

ASJC Scopus subject areas

  • General

Cite this

Feature selection based on statistical analysis. / Tahir, Nooritawati Md; Hussain, Aini; Abdul Samad, Salina; Husain, Hafizah; Jamaluddin, Mohd. Yusoff.

In: European Journal of Scientific Research, Vol. 14, No. 3, 2006, p. 426-433.

Research output: Contribution to journalArticle

Tahir, NM, Hussain, A, Abdul Samad, S, Husain, H & Jamaluddin, MY 2006, 'Feature selection based on statistical analysis', European Journal of Scientific Research, vol. 14, no. 3, pp. 426-433.
Tahir, Nooritawati Md ; Hussain, Aini ; Abdul Samad, Salina ; Husain, Hafizah ; Jamaluddin, Mohd. Yusoff. / Feature selection based on statistical analysis. In: European Journal of Scientific Research. 2006 ; Vol. 14, No. 3. pp. 426-433.
@article{2615d74b6c22489b96adc2cd2a6a0500,
title = "Feature selection based on statistical analysis",
abstract = "In most pattern recognition (PR) system, selecting the best feature vectors is an important task. Feature vectors serve as a reduced representation of the original data that facilitate us to evade the curse of dimensionality in a PR task. In this work, we deem further endeavor in selecting the best feature vectors for the PR task that is to determine the best eigenfeatures of four main human postures based on the rules of thumb of Principal Component Analysis namely the KG-rule, Cumulative Variance and the Scree Test. Accordingly, all three rules of thumb suggest in retaining only 9{\%} of the total eigenvectors or also known as 'eigenpostures'. Next, these eigenpostures are statistically analyzed prior to classification. Thus, the most relevant component of the selected eigenpostures can be ascertained. The statistical significance of the eigenpostures is determined using ANOVA. Further, a Multiple Comparison Procedure (MCP) and homogeneous subsets tests are performed to determine the number of optimized eigenpostures for classification. These optimized eigenpostures will feat as inputs to the Artificial Neural Network (ANN) classifier. The statistical analysis has enabled us to perform effectively the selection of eigenpostures for classification of human postures.",
author = "Tahir, {Nooritawati Md} and Aini Hussain and {Abdul Samad}, Salina and Hafizah Husain and Jamaluddin, {Mohd. Yusoff}",
year = "2006",
language = "English",
volume = "14",
pages = "426--433",
journal = "European Journal of Scientific Research",
issn = "1450-202X",
publisher = "European Journals Inc.",
number = "3",

}

TY - JOUR

T1 - Feature selection based on statistical analysis

AU - Tahir, Nooritawati Md

AU - Hussain, Aini

AU - Abdul Samad, Salina

AU - Husain, Hafizah

AU - Jamaluddin, Mohd. Yusoff

PY - 2006

Y1 - 2006

N2 - In most pattern recognition (PR) system, selecting the best feature vectors is an important task. Feature vectors serve as a reduced representation of the original data that facilitate us to evade the curse of dimensionality in a PR task. In this work, we deem further endeavor in selecting the best feature vectors for the PR task that is to determine the best eigenfeatures of four main human postures based on the rules of thumb of Principal Component Analysis namely the KG-rule, Cumulative Variance and the Scree Test. Accordingly, all three rules of thumb suggest in retaining only 9% of the total eigenvectors or also known as 'eigenpostures'. Next, these eigenpostures are statistically analyzed prior to classification. Thus, the most relevant component of the selected eigenpostures can be ascertained. The statistical significance of the eigenpostures is determined using ANOVA. Further, a Multiple Comparison Procedure (MCP) and homogeneous subsets tests are performed to determine the number of optimized eigenpostures for classification. These optimized eigenpostures will feat as inputs to the Artificial Neural Network (ANN) classifier. The statistical analysis has enabled us to perform effectively the selection of eigenpostures for classification of human postures.

AB - In most pattern recognition (PR) system, selecting the best feature vectors is an important task. Feature vectors serve as a reduced representation of the original data that facilitate us to evade the curse of dimensionality in a PR task. In this work, we deem further endeavor in selecting the best feature vectors for the PR task that is to determine the best eigenfeatures of four main human postures based on the rules of thumb of Principal Component Analysis namely the KG-rule, Cumulative Variance and the Scree Test. Accordingly, all three rules of thumb suggest in retaining only 9% of the total eigenvectors or also known as 'eigenpostures'. Next, these eigenpostures are statistically analyzed prior to classification. Thus, the most relevant component of the selected eigenpostures can be ascertained. The statistical significance of the eigenpostures is determined using ANOVA. Further, a Multiple Comparison Procedure (MCP) and homogeneous subsets tests are performed to determine the number of optimized eigenpostures for classification. These optimized eigenpostures will feat as inputs to the Artificial Neural Network (ANN) classifier. The statistical analysis has enabled us to perform effectively the selection of eigenpostures for classification of human postures.

UR - http://www.scopus.com/inward/record.url?scp=33749141253&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33749141253&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:33749141253

VL - 14

SP - 426

EP - 433

JO - European Journal of Scientific Research

JF - European Journal of Scientific Research

SN - 1450-202X

IS - 3

ER -