EcoSta 2018: Registration
View Submission - EcoSta2018
A0511
Title: Accelerated failure time modeling via continuous Gaussian scale mixtures Authors:  Sangwook Kang - Yonsei University (Korea, South) [presenting]
Byungtae Seo - Sungkyunkwan University (Korea, South)
Abstract: A semiparamtric accelerated failure time (AFT) model resembles the usual linear regression model - the response variable being the logarithm of failure times while the random error term is left unspecified. Thus, it is more flexible than parametric AFT models that assume parametric distributions for the random error term. Estimation for model parameters is typically done through a rank-based procedure, in which the intercept term cannot be estimated. This requires a separate estimation procedure for the intercept, which often leads to unstable estimates. For a better estimation of the intercept essential in estimating mean failure times or survival functions, we propose to employ a mixture model approach. To leave the model as flexible as possible, we consider nonparametric infinite scale mixtures of normal distributions. An expectation-maximization (EM) method is used to estimate model parameters. Finite sample properties of the proposed estimators are investigated via an extensive stimulation study. The proposed estimators are illustrated using a real data analysis.