Estimation of Change-Points Using Statistical Regression Stepwise Methods

• Main Authors: Tay Fang Wu, 吳泰芳
• Other Authors: Tze Fen Li
• Format: Others
• Language: zh-TW
• Published: 1999
• Online Access: http://ndltd.ncl.edu.tw/handle/04980461814353523097

碩士 === 國立中興大學 === 應用數學系 === 87 === In recent years, the statistical literature about the problem of the detection of change-points is really discussed widely. Now, we will explore a special and distinct form of the problem of detecting change-points. Here the Change-Point Model is similar to the general multiple regression model, and our main problem is to estimate and find the locations of change-points. The idea is that we assume that every point is a change-point, and we will use statistical regression stepwise methods to identify the true change-points and remove the false change-points. Therefore, by using statistical regression stepwise methods such as backward elimination procedure, forward selection procedure and stepwise regression procedure, we will obtain the quick and accurate estimates of the parameters of the change-point model, and the results of a Monte Carlo simulation study are presented to demonstrate the favorable estimation which is the final goal we want to accomplish.