|
Modeling on survival data , regression analysis is usually of a random variable and one or more controllable variables correlation. When data exist censoring , commonly used method is the Cox proportional hazards regression model , it is applied flexibly , to get attention. However, the Cox model must meet two assumptions , which sometimes does not hold two assumptions . This article uses another method - censored quantile regression , Cox model it as a supplement , the use of more flexible, especially its survival time for direct modeling, regression coefficients easier to interpret and to estimate the amount of asymptotic normality in nature. Also it can use a different sub- sites regression function to characterize the conditional distribution characteristics in different locations , for the analysis of survival data provides a natural and effective way . Censored quantile quantile as the promotion , the paper first introduces the quantile -related knowledge, the main content partakers digit calculation , nature and the estimated amount of asymptotic properties . Then discusses the basic theory of survival analysis , including Cox regression survival data in the application of censored quantile related content . Finally censored quantile applied to the treatment of diabetes survival data , in the survival time of the different sub- sites , we obtain a different regression function , so as to provide different levels of diabetes prevention and treatment measures reasonable .
|