DECISION TREE TO CLASSIFY JETS OF PARTICLES IN HIGH-ENERGY PHYSICS
No Thumbnail Available
Date
2025-04-10
Authors
Khaled, Marwa
Journal Title
Journal ISSN
Volume Title
Publisher
An-Najah National University
Abstract
This study investigates the application of Decision Tree and Random Forest models to tackle the challenge of jet classification in particle physics, focusing on categorizing jets into five key particle types: light quarks (q), gluons (g), W and Z bosons, and top quarks. Using a dataset from Zenodo comprising list of 53 High-Level Features derived from jet events, we explored Decision Tree and Random Forest models within the machine learning framework to enhance classification accuracy. Our approach involved building and comparing various Decision Tree and Random Forest models, assessing configurations such as tree depth, the minimum number of samples leaf, and the number of trees in the Random Forest ensemble to optimize performance.
In our jet classification research, the Random Forest model outperforms the Decision Tree model in classifying particle physics events, achieving higher precision, recall, F1-scores, and an overall accuracy of (85.32%) compared to (81.32%). Optimal Random Forest performance was obtained using 100 trees, maximum depth = 10, and top 20 selected features, which also reduced training time by 36.8%. In contrast, the best Decision Tree configuration used maximum depth = 8 and maximum features = 40. This research highlights the potential of Random Forest to achieve high jet classification accuracy and offers insights into optimizing Random Forest models for similar tasks in particle physics research.