Markov Processes for Big Data Hockey Analytics

点击翻译按钮获取中文摘要

Our project investigates big data analytics for ice hockey. It combines the most informative data available with the most powerful statistical model known from machine learning research to provide analytics for Canada's favorite game. The market for analytics based on big sports data is growing rapidly, with major international companies receiving millions of investment dollars. Canada has produced excellent research in sports analytics, but not much of the research expertise has been transferred to commercial products. The Montreal-based company SportLogiq uses advanced computer vision to extract from video extensive information about events in a hockey match. Their information is more detailed than that provided by any other company or organization. Our project will use machine learning to build a powerful statistical model for their data, known as a Markov process. Applications of statistical modeling include "Moneyball"-type ranking of player performance to find players that complement a team well.