作者: J. O'Quigley , P. Flandre
关键词:
摘要: A measure of the predictive capability a proportional hazards regression is derived. The based on residuals appropriate to regression. population version presented and can be seen not depend censoring mechanism under provision that any such independent or conditionally failure given covariate. For special case Weibull model, for which covariate distribution follows binary, uniform, normal, exponential laws, we derive analytic results. These alone give credence reflect strength effect, as quantified by parameter estimate, although scale between 0 1, independently intercept shape particular law only weakly dependent distribution. Extensions partial multiple measures ability are straightforward. An example provided.