| Author | Sumruay Komlayut |
| Call Number | AIT Thesis no. CS-91-7 |
| Subject(s) | Time-series analysis--Computer programs
|
| Note | A thesis submitted in partial fulfillment of the requirement for the degree of Master of Science, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Abstract | In time-series modelling, the selection of an appropriate model plays a very important role. For a good selection, several criteria have been proposed in addition to the portmanteau test. The main purpose of the present research is to carry out a simulation study in order to evaluate the performance of some important criteria, such as the Akaike Information Criterion (AIC), the new version of Akaike Information Criterion (AICI), Final Prediction Error (FPE), Posterior Possibility Criterion (PPC), etc .. A computer package which consists of two parts was developed. The first part that developed for using on a microcomputer deals with the preliminary analysis including plotting the data sequence, plotting the autocorrelation function (acf) and partial autocorrelation function (pacf). It also plots the residual series after the analysis together with the portmanteau test and test of randomness which is the diagnostic checking stage. The second part deals with the parameters estimation which is done on a SONY workstation. From the study, it was found that the AIC, AICI and FPE perform quite satisfactorily in determining the "best" model to represent the data set. Moreover, the package has been found to be very helpful in carrying all the tasks required in time-series analysis. |
| Year | 1991 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Other Field of Studies (No Department) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Huynh Ngoc Phien |
| Examination Committee(s) | Nagarur, Nagendra N. ;Jr. R.L. Gonzales |
| Scholarship Donor(s) | DADD |
| Degree | Thesis (M.Sc.) - Asian Institute of Technology, 1991 |