| Author | Bundit Soondharamanukit |
| Call Number | AIT Thesis no. CA-85-12 |
| Subject(s) | Estimation theory--Computer programs
|
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Abstract | Parameter Estimation for a mixed population has long been a subject of interest since the early days of statistical analysis. Despite a lot of theoretical work, practical success is still limited. At present, as computer speed increases, the method of maximum likelihood (ML), once considered impractical because of the amount of computation required, becomes more important. This thesis presents a practical solution based on ML. By using direct search and expectation-maximization (EM) procedure to search for the maximum on the likelihood surface, the flexibility is such that it can accommodate many families of continuous probability density functions, and with a mixture of two components, performance in terms of speed and "goodness of fit" is satisfactory. Small sample property (n=30) is also quite encouraging. |
| Year | 1985 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Other Field of Studies (No Department) |
| Academic Program/FoS | Computer Application (CA) |
| Chairperson(s) | Huynh, Ngoc Phien |
| Examination Committee(s) | Kanchit Malaivongs ;Kaew Naulchawee |
| Scholarship Donor(s) | Carl Duisburg Gesellschaft (CDG) |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1985 |