In this article, we provide an overview of well-known beta algorithms. We first study a stochastic search procedure proposed by Kennedy (1988) that asymptotically generates a beta variate. The goal is to identify the optimal parameter setting so that Kennedy`s algorithm can achieve the fastest speed of generation. For comparative purposes, we next evaluate the performance of some selected beta algorithms in terms of the following criteria: (i) validity of choice of shape parameters; (ii) computer generation time; (iii) initial set-up time; (iv) goodness of fit; and (v) amount of random number generation required. Based on the empirical study, we present three useful guidelines for choosing the best suited beta algorithm.
Relation:
Communications in Statistics - Simulation and Computation,38(4),750-770