Data-mining and expert models for predicting injury risk in ski resorts

Abstract

This paper proposes several models for predicting global daily injury risk in ski resorts. There are three types of models proposed: based on data mining, expert modelling, and a combination of both. We show that the expert model that represents the judgment of injury risk experts in the analyzed ski resort is 10–15% less accurate than data mining models. We also show that expert models refined with data-driven analysis can produce models that are in line with accuracy as data mining models, but in addition show some advantages, like transparency, consistency and completeness.