Djehiche, Boualem and KTH, Skolan för teknikvetenskap (SCI), Matematik (Inst.), Matematisk statistik
Springer Proceedings in Mathematics and Statistics. :127-147
Subjects
Natural Sciences, Mathematics, Probability Theory and Statistics, Naturvetenskap, Matematik, Sannolikhetsteori och statistik, Claims reserving, Disability insurance, Life insurance, Mortality modeling, Thiele’s equation, Estimation, Health insurance, Insurance, Mortality model, Statistical estimation, and Statistical methods
Dongen, Boudewijn van, Carmona Vargas, Josep, Chatain, Thomas, Universitat Politècnica de Catalunya. Departament de Ciències de la Computació, and Universitat Politècnica de Catalunya. ALBCOM - Algorismia, Bioinformàtica, Complexitat i Mètodes Formals
Subjects
Àrees temàtiques de la UPC::Informàtica::Sistemes d'informació, Data mining, Enterprise resource management, Management science, Statistical methods, In-process, Leave-one-out cross validations, Measure precision, Measuring precision, Process model, Unified approach, and Mineria de dades
Abstract
The holy grail in process mining is an algorithm that, given an event log, produces fitting, precise, properly generalizing and simple process models. While there is consensus on the existence of solid metrics for fitness and simplicity, current metrics for precision and generalization have important flaws, which hamper their applicability in a general setting. In this paper, a novel approach to measure precision and generalization is presented, which relies on the notion of antialignments. An anti-alignment describes highly deviating model traces with respect to observed behavior. We propose metrics for precision and generalization that resemble the leave-one-out cross-validation techniques, where individual traces of the log are removed and the computed anti-alignment assess the model’s capability to describe precisely or generalize the observed behavior. The metrics have been implemented in ProM and tested on several examples.