Geographically Weighted Regression in Cox Survival Analysis for Weibull Distributed Data with Bayesian Approach

9th Annual Basic Science International Conference 2019

自然科学（总论）

Taufiq, Ahmad^1 ; Astuti, Ani Budi^1 ; Rinaldo Fernandes, Adji Achmad^1

Department of Statistics, Brawijaya University, Jl Veteran, Malang, Indonesia^1

关键词: Bayesian approaches;    Data distribution;    Distributed data;    Geographically weighted regression;    Posterior distributions;    Prior distribution;    Survival analysis;    Survival Function;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/546/5/052078/pdf
DOI  :  10.1088/1757-899X/546/5/052078

学科分类：自然科学（综合）

来源: IOP

【 摘 要 】

Cox survival analysis is a statistical method used in survival data, which examines an event or occurrence of a particular event. In survival analysis, the response variable is survival time, usually called the T failure event. In the development, survival analysis involves spatial effects. One of the spatial effects is point effect, which coordinates of adjacent points will give an influence. Spatial model involves points called Geographically Weighted Regression (GWR). In this research, data distribution used is Weibull distribution, which survival time data is divided into three periods. Parameter estimation used is Bayesian Approach. Bayesian approach is better used in survival analysis that has a lot of censored data. The research purpose is getting survival function, hazard function, and Cox survival model with GWR and Weilbull distributed data and determining the prior distribution and posterior distribution in Bayesian approach. The result of this research is reducing the new hazard function from Weibull distribution and changing μ to become the GWR model, and then obtained model is . Parameters in the result are estimated using Bayesian approach.