The estimation method for clustered failure time data
碩士 === 國立臺北大學 === 統計學系 === 106 === Joint likelihood estimation and pairwise likelihood estimation for clustered data in multivariate failure time (MVFT) are evaluated, and data are assumed to have a characteristic of competing risks. This study will consider two or three observations per cluster and...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2018
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/nez45q |
| Summary: | 碩士 === 國立臺北大學 === 統計學系 === 106 === Joint likelihood estimation and pairwise likelihood estimation for clustered data in multivariate failure time (MVFT) are evaluated, and data are assumed to have a characteristic of competing risks. This study will consider two or three observations per cluster and use Weibull distribution as the marginal distribution and different Copula to construct a joint survival distribution. Using statistical software to simulate data, parameter estimation is performed. The main estimated parameters are the parameters in the marginal distribution and the dependence parameter in the Copula. The simulation results of the joint likelihood estimation are well, and they have good performance on the evaluation index; the pairwise likelihood estimation is not stable enough, and only meets the evaluation criteria in some cases.
|
|---|