Development of Inference Methods for Finite Normal Mixture Models
| Project/Area Number |
26380267
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Economic statistics
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
SHIMOTSU Katsumi 東京大学, 大学院経済学研究科(経済学部), 教授 (50547510)
|
|---|
| Project Period (FY) |
2014-04-01 – 2017-03-31
|
| Project Status |
Completed (Fiscal Year 2016)
|
| Budget Amount *help |
¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
Fiscal Year 2016: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2015: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Fiscal Year 2014: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
|
|---|
| Keywords | 正規有限混合モデル / 計量経済学 / 統計的推測 / 漸近理論 / 有限混合モデル / 正規混合回帰モデル |
|---|
| Outline of Final Research Achievements |
This research project developed new inference methods for finite mixture of normal regression models (switching regression models) and multivariate normal mixture models. These mixture models have been used in numerous empirical applications in economics. The number of components is an important element of these mixture models. However, because the log-likelihood function of a normal mixture model has a non-standard structure, there exists no procedure to statistically test the number of components in these models. We develop a new approach to analyzing the log-likelihood function of these two models, and construct likelihood-based tests for the number of components. Further, we confirmed by computer simulations that the proposed tests show good finite sample performance.
|
Report
(4 results)
Research Products
(11 results)
