Presentation 2007-12-22
A Study on Conditional Quantile Estimation for Location-Scale Models
Mitsuhiro YOSHIDA, Yusuke HIGUCHI, Ichiro TAKEUCHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A family of conditional quantile functions provides deep insight into the underlying stochastic relationship between variables. However, when several conditional quantile functions at different orders (e.g., 0.1, 0.2, …, 0.9 quantiles) are independently estimated, their inter-relationship could be inconsistent. To address this problem, we consider estimating a set of conditional quantile functions under an assumption that the data is drawn from location-scale model. We study the inter-relationship between different conditional quantile functions under location-scale model, and introduce a new approach for conditional quantile estimations. For linear models, we suggest a computationally efficient approach using linear programming, and for nonlinear models, we introduce a multi-layer perceptron with special structure. Numerical experiments suggest the effectiveness of our new approach.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) location scale model / quantile regression / linear programing / multi-layer perceptron
Paper # NC2007-71
Date of Issue

Conference Information
Committee NC
Conference Date 2007/12/15(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 Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Study on Conditional Quantile Estimation for Location-Scale Models
Sub Title (in English)
Keyword(1) location scale model
Keyword(2) quantile regression
Keyword(3) linear programing
Keyword(4) multi-layer perceptron
1st Author's Name Mitsuhiro YOSHIDA
1st Author's Affiliation Information Engineering Division, Department of Engineering, Mie University()
2nd Author's Name Yusuke HIGUCHI
2nd Author's Affiliation Information Engineering Division, Department of Engineering, Mie University
3rd Author's Name Ichiro TAKEUCHI
3rd Author's Affiliation Information Engineering Division, Department of Engineering, Mie University
Date 2007-12-22
Paper # NC2007-71
Volume (vol) vol.107
Number (no) 410
Page pp.pp.-
#Pages 6
Date of Issue