Presentation 2014-11-17
Approximate Models of Probability Distributions for Principal Components of Sample Mahalanobis Distances : About Each Element and Partial Sum of Sample Mahalanobis Distances
Yasuyuki KOBAYASHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Probability distributions of the principal components and their partial sum, into which a sample Mahalanobis distance is decomposed by the eigenvalues and eigenvectors of the sample covariance matrix, are said to follow F distributions. However, the distributions do not follow F distributions when the number of the learning sample is small. This report has proposed approximate models of the distributions with x^2 or gamma distributions by analysis of the delta method in mathematical statistics, and has obtained more coincident results with the approximate models than with F distributions by numerical experiments.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Mahalanobis distance / principal components / partial sum / probability distribution / approximate models
Paper # IBISML2014-37
Date of Issue

Conference Information
Committee IBISML
Conference Date 2014/11/10(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Information-Based Induction Sciences and Machine Learning (IBISML)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Approximate Models of Probability Distributions for Principal Components of Sample Mahalanobis Distances : About Each Element and Partial Sum of Sample Mahalanobis Distances
Sub Title (in English)
Keyword(1) Mahalanobis distance
Keyword(2) principal components
Keyword(3) partial sum
Keyword(4) probability distribution
Keyword(5) approximate models
1st Author's Name Yasuyuki KOBAYASHI
1st Author's Affiliation Faculty of Science and Engineering, Teikyo University()
Date 2014-11-17
Paper # IBISML2014-37
Volume (vol) vol.114
Number (no) 306
Page pp.pp.-
#Pages 8
Date of Issue