Model-based quantile regression for count panel data
Panel data are observed in many research areas such as econometrics, social sciences and medicine. It involves repeated observations of the same subjects over a short or long period of time, where the multiple subjects are independent but the repeated measurements over time within one subject are no...
Saved in:
| 主要作者: | |
|---|---|
| 其他作者: | |
| 格式: | Final Year Project |
| 語言: | English |
| 出版: |
2019
|
| 主題: | |
| 在線閱讀: | http://hdl.handle.net/10356/77146 |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 總結: | Panel data are observed in many research areas such as econometrics, social sciences and medicine. It involves repeated observations of the same subjects over a short or long period of time, where the multiple subjects are independent but the repeated measurements over time within one subject are non-independent.
The objective of the Final Year Project (FYP) is to propose a model-based quantile regression method to estimate count panel data. By linking Generalized Linear Mixed Model (GLMM) based on Poisson distribution and the Quantile Regression (QR) model, we can map the parameters of the response variable to the regression quantiles and then estimate the regression quantiles through the likelihood function with Asymmetric Laplace Distribution (ALD). On top of that, an extension of the discrete responses is explored by adding continuous generalization to the response variable. |
|---|