FORECASTING PLAYING STYLE OF A FOOTBALL TEAM WITH ARIMA MODEL

• Main Author: ZEN (NIM : 13511060), MUHAMMAD
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/23352
• Institution: Institut Teknologi Bandung

Football is a sport that heavily involves strategies. Each of the teams has their own playing style as a strategy to beat their opponents. In a football match, there are a lot data from the events happening during the match. Those data can be used for forecasting playing style of a football team. If the playing style of a team is known, the strategy for beating that team can be prepared. One of models that can be used for forecasting is Autoregresif Integrated Moving Average (ARIMA) model. This model is often used for forecasting stocks. The training data used for forecasting playing style of a football team with ARIMA model are match data of Arsenal versus Chelsea from 2014/2015 season to 2016/2017 season. The variables used to represent the playing style of a football team are goals, shots, shots on target, shots off target, ball possessions, offsides, fouls, corner kicks, opponent’s shots, opponent’s shots on target, opponent’s shots off target, opponent’s offsides, fouls opponent committed, opponent’s corner kicks, yellow and red cards. To calculate the performance of this model, mean squared error is used. Based on prediction with ARIMA model the error of goals predictions are 0.5, the error of shots predictions are 25, the error of shots on target predictions are 9, the error of shots off target predictions are 90, the error of ball possessions predictions are 134.5, the error of offsides predictions are 2.5, the error of fouls predictions are 298, the error of corner kicks predictions are 58, the error of opponent’s shots predictions are 169, the error of opponent’s shots on target predictions are 25, the error of opponent’s shots off target predictions are 64, the error of opponent’s offsides predictions are 2, the error of fouls opponent committed predictions are 143.333, the error of opponent’s corner kicks predictions are 1, the error of yellow cards predictions are 1, the error of red cards predictions are 0.