Details of Research Outputs

In this paper, we present HyperStar2, a tool for fittingMarkov Arrival Processes (MAPs) to empirical data. HyperStar2 uses a two-step approach, where the first step is clusterbased fitting of phase-type distributions and the second step is the construction of a correlation matrix. In the first step, we use the cluster-based algorithm for Hyper-Erlang distribution fitting from HyperStar [1]. Based on the Hyper-Erlang fitting result and the clusters of samples, in the second step we construct the correlation matrix. The tool targets engineers and scientists who need distribution fitting for non-standard distributions but have little interest in the underlying theory. Therefore the tool has a GUI that offers graphical presentation of the data, the fitted distribution and the empirical as well as theoretical autocorrelation. We discuss the use of HyperStar2 in common application scenarios and the fitting algorithms behind it. We provide some numerical examples, which show the abilities and limits of the fitting tool. We find that HyperStar2 can fit distributions very well but for some autocorrelation structures ProFiDo provides better results for autocorrelation.