Discrete Time Models

Abstract

As noted in section 2.3, by the introduction of failure indicators ${{y}_{ij}},\,i=1,\ldots ,n,j=1,\ldots ,{{t}_{i}}$ a Bernoulli likelihood is obtained and estimation can proceed as for binary regression – allowing that time can be treated like an arbitrary covariate whose effect can be smoothed. This is not the case for continuous time models.