Kernel-based Estimation Method

Zhang, D. (2008). Kernel-based Estimation Method. IIASA Interim Report. IIASA, Laxenburg, Austria: IR-08-025

[thumbnail of IR-08-025.pdf]
Preview
Text
IR-08-025.pdf

Download (417kB) | Preview

Abstract

Regression is a basic statistical tool for estimation task of data mining, which is to predict the relationship between a dependent variable and one or more independent variables. Parametric and nonparametric regressions are two kinds of regression approach used for various problems. This work proposes a kernel-based nonparametric regression method, which can solve nonlinear regression problem properly by mapping the data to a higher-dimensional space by kernel function. With this method, we conduct a series of experiment on nonlinear function and real world regression problems, and the results reveal the effectiveness of the model. The results reveal that the model is efficient on some data sets with similar or even higher precision than the prevalently used support vector regression and neural network regression method. Nevertheless, there are still other data sets which kernel-based method cannot works well, as water flow and forest fire data set.

Item Type: Monograph (IIASA Interim Report)
Research Programs: Integrated Modeling Environment (IME)
Young Scientists Summer Program (YSSP)
Depositing User: IIASA Import
Date Deposited: 15 Jan 2016 08:41
Last Modified: 27 Aug 2021 17:20
URI: https://pure.iiasa.ac.at/8759

Actions (login required)

View Item View Item