以多維負二項模型分析具相關性的混合型態資料

Multivariate negative binomial models for correlated data of mixing types

朱美穎(Mei-ying Chu)

指導教授 ： 鄒宗山

國立中央大學/理學院/統計研究所/碩士(2011年)

摘要

具相關性的連續與個數資料，如一個人的身高、體重與血壓值，常見於醫學與其他研究領域中。這種相關性的資料在分析上較困難，原因在不易找到適合的統計模型。
本文中透過不同實作模型的比較，來說明強韌多維負二項模型是個可適用於任何資料類型的統計模型。

關鍵字

強韌概似函數 ； 混合模型

並列摘要

Multivariate correlated data with various types are encountered in many research areas. Such data are generally more difficult to analyze due to the scarcity of appropriate statistic models.
We demonstrate that the multivariate negative binomial model can be easily modified to provide asymptotically legitimate inferences about the regression parameters for correlated data of mixing types. Contrast is also made with models proposed in the literatures.

並列關鍵字

robust likelihood function ； mixture model

參考文獻

Arbous, A. G. and Kerrich, J. E. (1951). Accident statistics and the concept of

accident proneness , Biometrics, 7, 340-432.

Kang, J., Mao, K., Yang, Y., and Zhang, J. (2007). Regression models for mixed Poisson and continuous longitudinal data, Statistics in Medicine, 26, 3782-3800.

Royall, R. and Tsou, T. S. (2003). Interpreting statistical evidence by using imperfect models: robust adjusted likelihood functions, Journal of the Royal Statistical Society, Series B, 65, 391-404.

Solis-Trapala, I. L. and Farewell, V. T. (2005). Regression analysis of overdispersed correlated count data with subject specific covariates. Statistics in Medicine, 24, 2557-2575.