PREDICTION OF ENGLISH PREMIER LEAGUE 2018-2019 SEASON FINAL TABLE USING BAYESIAN HIERARCHIAL AND MARKOV CHAIN MONTE CARLO

• Main Author: Dea Sukmana, Fathurobbi
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/38683
• Institution: Institut Teknologi Bandung

Football is a famous sport in this earth. Almost all country in this earth have a football league on their country. One of the famous football leagues is English Premier League, a football league in England. In the recent few years, English Premier League became a fairly fierce league and there are a lot of unpredictable full-time score happened on a match on English Premier League. In this research, A Bayesian Hierarchial model is created, and used for predict the full-time score of a match on English Premier League, by compare the sum of goal(s) scored by the two team that battle on the match. The model is influenced by some factor that influence a goal happened, such as attack power of a team, defense power of a team, external factor and a home factor, a factor of plays a match in a home. In the beginning, all those factors generated randomly by using a Metropolis-Hasting algorithm that uses the reversibility principle of Markov Chain Monte Carlo, then some best sample that maximize the posterior distribution of each factor are chose by using Maximum A Posteriori method. After gaining all those factors, prediction of a full-time score of a match could be done for a full-season, so that the champion of English Premier League 2018-2019 season can be predicted.