Extracting decision models for ski injury prediction from data

Creating decision models for risk assessment of ski injuries is a challenging task. Ski are rare events, but they carry high cost, that is, can cause working or movement disabilities. Usually, performed on small-scale, case-controlled studies where the effect single factor evaluated. Recently, data mining and machine learning algorithms being employed injury prediction. However, these do not generally satisfy need interpretation model, provide explanations predictions, in general ensure completeness consistency rules. To make useful, one needs to implement aforementioned properties. Decision support systems expected have properties; however, process building such still tedious: it has consider human biases, assumptions, subjective values, as well focus problem solved. We propose method extraction from at hand. Our DIDEX, Data Induced DEcision eXpert, builds desirable properties inclusion systems. The proposed used build model prediction based Mt. Kopaonik resort, Serbia. results show DIDEX generates up five times simpler compared existing domain expert DEX while having 6% better predictive accuracy. Additionally, its accuracy comparable similar algorithms, tree classifiers, random forest, logistic regression.