mailto:uumlib@uum.edu.my 24x7 Service; AnyTime; AnyWhere

Robust parameter estimation method for bilinear model

Ismail, Mohd Isfahani and Ali, Hazlina and Syed Yahaya, Sharipah Soaad (2015) Robust parameter estimation method for bilinear model. AIP Conference Proceedings, 1691 (1). 050012. ISSN 0094-243X

Full text not available from this repository. (Request a copy)

Abstract

This paper proposed the method of parameter estimation for bilinear model, especially on BL(1,0,1,1) model without and with the presence of additive outlier (AO). In this study, the estimated parameters for BL(1,0,1,1) model are using nonlinear least squares (LS) method and also through robust approaches.The LS method employs the Newton-Raphson (NR) iterative procedure in estimating the parameters of bilinear model, but, using LS in estimating the parameters can be affected with the occurrence of outliers. As a solution, this study proposed robust approaches in dealing with the problem of outliers specifically on AO in BL(1,0,1,1) model.In robust estimation method, for improvement, we proposed to modify the NR procedure with robust scale estimators. We introduced two robust scale estimators namely median absolute deviation (MADn) and Tn in linear autoregressive model, AR(1) that be adequate and suitable for bilinear BL(1,0,1,1) model. We used the estimated parameter value in AR(1) model as an initial value in estimating the parameter values of BL(1,0,1,1) model.The investigation of the performance of LS and robust estimation methods in estimating the coefficients of BL(1,0,1,1) model is carried out through simulation study. The achievement of performance for both methods will be assessed in terms of bias values.Numerical results present that, the robust estimation method performs better than LS method in estimating the parameters without and with the presence of AO.

Item Type: Article
Additional Information: Presented at the 2nd Innovation and Analytics Conference & Exhibition (IACE 2015), 29 September–1 October 2015, Kedah, Malaysia
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: School of Quantitative Sciences
Depositing User: Dr. Hazlina Ali
Date Deposited: 16 Apr 2017 02:32
Last Modified: 16 Apr 2017 02:32
URI: https://repo.uum.edu.my/id/eprint/21570

Actions (login required)

View Item View Item