Hyperparameter Optimized KNN Models for Recommendation Systems

Udatha, Harishbabu (2024) Hyperparameter Optimized KNN Models for Recommendation Systems. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration manual) Download (742kB) | Preview

Abstract

Recommendation systems are essential to improving user experiences since they offer tailored recommendations. This work investigates the use of machine learning approaches to enhance movie recommendation systems, with an emphasis on content- based and collaborative filtering strategies. This study examines how to optimise k- nearest-neighbors (KNN) models using content-based filtering using term frequency- inverse document frequency (TF-IDF) and hyperparameter optimisation with Optuna using the MovieLens dataset. The motivation is the need to increase the precision of movie recommendations while taking into account a variety of user preferences. By combining Optuna for hyperparameter tuning and TF-IDF for feature optimisation, our research seeks to close the recommendation accuracy gap. Our unique optimisation methods for KNN models for content-based and collaborative filtering make these contributions. Extensive analyses show that the recommendations are more accurate, and the accuracy has improved significantly overall. Our work puts recommendation systems in line with state-of-the-art techniques by demonstrating, theoretically, the efficacy of complex optimisation techniques. Practically speaking, the key benefit is providing users with more personalised and accurate movie recommendations, which improves their viewing experiences in general. Content-based filtering showed a moderate level of accuracy with an MSE of 7.7652, with notable differences between recorded and expected ratings. Collaborative filtering performed remarkably well with 15 neighbours and Pearson similarity, yielding a 0.0012 RMSE and very accurate user rating predictions.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Jain, Mayank UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Tamara Malone
Date Deposited:	07 Apr 2025 10:48
Last Modified:	07 Apr 2025 10:48
URI:	https://norma.ncirl.ie/id/eprint/7376

Actions (login required)

View Item