Comparison of contact distribution models with a stochastic ordering approach

Maria Kamińska-Zabierowska, Maciej Kamiński

doi:10.5277/arc160306

Abstract

No universal method for comparing contact models is known. Models we met in literature are usually stated by a complicated mathematical formula, which does not provide any information about the model’s specific behaviour, although it characterizes the model. In this paper the authors present their own way of comparing contact models. We run models in analytic theory on ideal theoretical distributions. Thanks to this we are able to compare multiple models and their behaviour in terms of arriving distribution or short/long trips preferences. We use stochastic ordering approach for fractions of completed contacts as the criterion for comparing models. The presented idea seems to be a good alternative to a comparison based on the model’s output for fixed city instances. Comparison is made for production constrained and intervening opportunity models. The results for chosen distributions show that, even if both models differ much in their assumptions, they can give very similar outputs.