|
Abstract:
|
In 1996, Hastie used pseudo-eigenvalue decomposition to approximate the smoothing matrix in an additive model setting based on orthonormal basis functions. In developing a new model, we extended the pseudospline approach to the multiplicative setting proposed by Hastie and Tibshirani (1990) and expanded on Cox's 1972 proportional hazards model. The use of pseudosplines in models of survival data offers several advantages, including flexibility in the specification of the form of the hazard function, a reduction in the computational burden of the estimation imposed by Gray's 1992 model of survival analysis, and simplification of model inferences. We will describe our pseudospline-based model and examine its results when we applied it to the analysis of survival data.
|
