Предвиђање броја скијашких повреда коришћењем алгоритама временских серија и машинског учења

Abstract

Skiing is a very popular sport, due to the possibility of high-speed development and frequent performance of acrobatic elements by skiers, there is a risk of serious injuries. To improve the decision-making process, keeping records and predicting ski injuries has become very important for ski resorts. This paper illustrates the application and comparison of the time series model and classical machine learning models in predicting the number of ski injuries on Kopaonik. The number of injuries is predicted on a daily, weekly and monthly basis. Predicting injuries at different time frequencies provides a better insight into the overall movement of the number of injuries over time, which certainly affects the quality of planning and decision making. The entire research was conducted in the Python programming language.