| Summary: | 碩士 === 國立雲林科技大學 === 工業工程與管理研究所碩士班 === 90 === Key word: bezier surface，regression，nonparametric，,consistency，nonlinear
A Robust Regression Analysis Using Bezier Surface
Abstract
It is an important subject to discuss the relation between two or three variables,In fact it has been applied in many fields for example in manufacture process, heuristic algorithm ,economic, etc.
Regression modal has been wildly used as analysis tool in industrial engineering and management, statistics, quality management. ,social science .The main purpose of decision is to reduce the risk and cost. It is important that how to model the regression in different situation . And make the analysis more precise and meaningful. and applied the model develop method in many area. In tradition linearly regression it is necessary to be proficient in the topic in order to make hypothesize.
But it is hard to access this approach. .In many situation the model is hard to hypothesize .especially in multiple variables .To overcome the difficulty ,approaches that are independent of an assumed model have been develop.. Hence we present a new regression analysis that use Bezier surface. Geometric model. this approach estimates the regression function by bezier surface without the need of a hypothesized model .To estimates the regression function was controlled only by the observations. It is the important property of the geometric model by converge .In statistics we prove that when the observations number is large the estimate model will converge to the true function regardless of linearly or nonlinearly.But in this study it is important to get the roots of the nolinealy eqution system.In order to get the solution efficiently.
We set the error of the X and Y Bezier eqution system.by0.05.Hence make the error of ture solution between Zd and Z widely.So each of the result in our experience will just offer the direction and reference.to understand the converge property of MSE.in our regression model of Bezier surface
|
|---|
