IMPLEMENTATION OF BAYESIAN NONPARAMETRIC-INFERENCE BASED ON BETA-PROCESSES

Hjort (1990) constructs prior distributions for cumulative hazards usi ng stochastic processes with non-negative independent increments, A pa rticular class of processes termed beta processes is introduced there for this purpose, It is also shown that the posterior cumulative hazar d is again a beta process given exact and right censored data, In this paper, we develop an algorithm that enables approximate random variat e generation from the posterior process using the Levy formula for its moment generating function, Computation is illustrated using failure time data.