Optimizing Recommendation Systems By Fusion of KNN, Singular Value Decomposition, and XGBoost for Enhanced Performance

Recommender systems are crucial in managing information overload by providing personalized suggestions based on user preferences. Traditional collaborative filtering methods like K-Nearest Neighbors (KNN) and Singular Value Decomposition (SVD) are effective but face sparse datasets and scalability challenges. This research introduces a hybrid collaborative filtering recommendation a lgorithm that integrates KNN, SVD Decomposition, and XGBoost, a gradient-boosting framework. This approach aims to enhance recommendation quality by combining KNN's neighborhood-based filtering, SVD's latent factor modeling, and XGBoost's predictive power. This research investigates the impact of different KNN similarity measurements on the accuracy of the XGBoost model within the hybrid framework. The goal is to identify the optimal similarity measurement that, when combined with SVD and XGBoost, results in accurate and personalized recommendations across various scenarios. Detailed analysis and experiments on benchmark datasets evaluate the model's performance in terms of accuracy, scalability, and computational efficiency. The results demonstrate that the hybrid model, enhanced with diverse KNN similarity measures, consistently outperforms standalone techniques and conventional ensemble methods. This research underscores the potential of the proposed model to develop more effective and scalable recommendation systems across diverse domains.