Linear Transformation Models For Survival Data Applied In Stock Market

We consider a class of semi-parametric linear transformation models, underwhich an unknown transformation of the survival time is linear related to thecovariates with various error distribution, which is known or unknown, that isg(T) = -β'Z +εwhere g(·) is assumed to be a smooth, invertible and strictly monotonically in-creasing function, Z is a p×1 covariate,βis a p×1 coeffcient vector, andεisan error. We consider the case in whichεfollows a known distribution, such asextreme value distribution. Our aim is to obtain the estimation ofβ.We discuss the linear transformation models for survival data, the inferencesare essentially based on the likelihood function, and the inference for the monotonetransformation is after the estimation forβ. In this paper, we propose a di?erentapproach: an estimator of the monotone transformation is given firstly, then basedon a transformation of the observed data, an estimator of coe?cients for covariatesis derived from the ordinary linear least squares procedure and the large sampleproperties are also obtained.In recent years, people have made many scientific researches on stoke market,and more discussion about yield, but few people applied survival analysis to stokemarket. We apply survival models to the yield of stoke, and regard successive risesand falls as a typical survival process. We use semi-parametric linear transforma-tion models to analysis this process, and then derive the relation of successive rises(falls) and volumes.