Serial Correlation Estimation by Multiple Objective Piecewise Regrsssion

• Main Authors: Chien-Wei Lee, 李建瑋
• Other Authors: Jing-Rung Yu
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/05096449753363647322

碩士 === 國立暨南國際大學 === 資訊管理學系 === 95 === This paper proposed MOGP/SC (Multiple Objective Goal Programming/Serial Correlation) model which could deal with the imprecise dependent and independent variables. The conventional GP/SC (Goal Programming/Serial Correlation) model was used as a time series analysis model with only for the precise dependent and independent variables. This paper has four properties: (1) Dealing with outlier data in piecewise regression by using quadratic programming; (2) Dealing with outlier data in piecewise regression by using GP/SC model; (3) Combining two objectives to minimize the number of change points and the deviation of the regression model simultaneously; (4) Dealing with the different data type of the dependent and independent variables. The aim of this paper is to develop a new formulation of GP model for regression with Serial Correlation where the dependent and independent variables are imprecise. By utilizing the piecewise concept, the proposed model can deal with outliers by automatically segmenting the data. We generate some outliers in our data sets and four examples are demonstrated to show our proposed method in more details.