IMPLEMENTATION OF EXPECTED GOAL MODEL USING GRADIENT BOOSTING ALGORITHM
Indonesia is one of the countries with huge football fan numbers. However, its achievement still fell behind other countries due to lack of technology usage in football quality development. Match analysis is one of the most effective methods on how a football team evaluates its own game while stu...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/79179 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Indonesia is one of the countries with huge football fan numbers. However, its achievement
still fell behind other countries due to lack of technology usage in football quality
development. Match analysis is one of the most effective methods on how a football team
evaluates its own game while studying the opponent's playing patterns. Expected Goal (xG)
is an indicator used to calculate the chance of a shot resulting in a goal. In producing the
expected goal model, machine learning based on supervised learning with the binary
classification type will be used. This research aims to produce the best expected goal
calculation model among three gradient boosting algorithms, namely CatBoost, XGBoost,
and LightGBM. Based on testing with Log Loss, Brier score, and AUROC score metrics, the
most accurate gradient boosting algorithm is XGBoost compared to another algorithm. |
|---|