Title: Exploiting advanced machine learning techniques for predictive analysis of novice learners' programming performance

Authors: Kapil Shukla; Parag Shukla

Addresses: School of Forensic Science, National Forensic Sciences University, Gujarat, India ' School of Computing, Kaushalya – The Skill University, Gujarat, India

Abstract: Programming education is evolving quickly, thus new methods are needed to help beginners learn and grow. This study predicts novice learners' performance utilising sophisticated machine learning methods including K-nearest neighbours, decision tree, random forest, and XGBoost. We assessed these models on accuracy, precision, recall, and F1-score utilising 2,111 samples, 11 beginning features, and derived attributes including correctness, error, performance, and final choice. Ensemble models like random forest and XGBoost capture complex data patterns better since they generalise robustly. Simple KNN and decision tree ensembles provide a foundation but have weak feature interactions and class distributions. Performance and prediction are improved via hyperparameter adjustment and feature engineering in this research. This research personalises/adapts novice learners' learning aids using predictive models. Educational data mining is growing, and machine learning may revolutionise programming education. This dataset may be expanded, environmental variables researched, or improved using deep learning.

Keywords: machine learning; ML; performance prediction; supervised learning algorithms; novice learner programming performance.

DOI: 10.1504/IJIL.2026.151467

International Journal of Innovation and Learning, 2026 Vol.39 No.2, pp.242 - 255

Received: 05 Dec 2024
Accepted: 08 Feb 2025
Published online: 02 Feb 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article