Date of Award
1-1-2009
Thesis Type
masters
Document Type
Thesis
Divisions
science
Department
Faculty of Science
Institution
University of Malaya
Abstract
The class of random coe±cient autoregressive (RCA) models has been con-sidered in many areas of science due to its rich applications. We review two methods of RCA parameter estimation, namely least squares and estimating functions. An iterative method based on the estimating functions is proposed to improve the existing RCA parameter estimation. This study is then fol-lowed by investigating the robustness of the three estimates when outliers exist in the RCA process. Simulation studies are carried out to investigate the per-formance of parameter estimation and robustness of the estimates.Further, the outlier detection procedure for the RCA process is proposed.In this study, a procedure by Chang et al. (1988) has been extended to detect additive and innovational outliers in the RCA process. A simulation study is carried out to investigate the performance of the procedures. It is found that, in general, these procedures work well in detecting outliers. Finally, we apply the suggested procedures to a real data set to show the importance of the study in practice.
Note
Dissertation (M.Sc.) -- Institut Sains Matematik, Fakulti Sains, Universiti Malaya, 2009
Recommended Citation
Abdullah, Norli Anida, "Estimation and outlier detection of random coefficient autoregressive models / Norli Anida binti Abdullah" (2009). Student Works (2000-2009). 2941.
https://knova.um.edu.my/student_works_2000s/2941