Mapfit: An R-based tool for PH/MAP parameter estimation

Abstract

In this paper, we present a tool for estimating parameters of phase-type distribution (PH) and Markovian arrival process (MAP) on the statistical analysis package R. PH and MAP are useful for the analysis of non-Markovian models approximately. By approximating general distributions and point processes with PH and MAP, the non-Markovian models is reduced to continuous-time Markov chains (CTMCs) that can be solved by analytical approaches. The significant features of our tool are (i) PH/MAP fitting from grouped data and (ii) PH fitting from probability density functions.