Differences

This shows you the differences between two versions of the page.

Link to this comparison view

blogs:pub2016:a_unified_approach_for_measuring_precision_and_generalization_based_on_anti-alignments [2016/10/04 11:03]
hverbeek
blogs:pub2016:a_unified_approach_for_measuring_precision_and_generalization_based_on_anti-alignments [2016/10/04 11:04] (current)
hverbeek
Line 3: Line 3:
 B.F. van Dongen, J. Carmona, and T. Chatain. [[http://​dx.doi.org/​10.1007/​978-3-319-45348-4_3|A Unified Approach for Measuring Precision and Generalization Based on Anti-Alignments]]. In La Rosa, Marcello and Loos, Peter and Pastor, Oscar (editors), [[http://​dx.doi.org/​10.1007/​978-3-319-45348-4|Business Process Management: 14th International Conference, BPM 2016]], Rio de Janeiro, Brazil, September 18-22, 2016. Proceedings,​ 2016, pages 39-56. ​ B.F. van Dongen, J. Carmona, and T. Chatain. [[http://​dx.doi.org/​10.1007/​978-3-319-45348-4_3|A Unified Approach for Measuring Precision and Generalization Based on Anti-Alignments]]. In La Rosa, Marcello and Loos, Peter and Pastor, Oscar (editors), [[http://​dx.doi.org/​10.1007/​978-3-319-45348-4|Business Process Management: 14th International Conference, BPM 2016]], Rio de Janeiro, Brazil, September 18-22, 2016. Proceedings,​ 2016, pages 39-56. ​
  
 +===== 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
 +anti-alignments. 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.
 +
 +===== Links =====
 +
 +  * {{:​blogs:​pub2016:​a_unified_approach_for_measuring_precision_and_generalization_based_on_anti-alignments.pdf|Preprint}}