Enhancing E-commerce Recommendation Accuracy Using KNN and Hybrid Approaches: An Empirical Study

In the contemporary e-commerce landscape, personalized recommendation systems play a vital role in enhancing the shopping experience. This study aimed to investigate the utility of the K-nearest neighbors (KNNs) algorithm in e-commerce recommendation systems and explore the potential benefits of integrating KNNs with other algorithms in a hybrid system. Using a publicly available dataset, we developed a KNN-based recommendation system and a hybrid system and evaluated their performance based on precision@k, recall@k, and F1 score metrics. Our results indicated that while the KNN-based system performed well, the hybrid system outperformed it in all evaluated metrics. This finding suggests the potential of combining KNN with other algorithms to improve recommendation performance. Future research directions include exploring advanced techniques for handling sparse data, addressing the cold-start problem, and experimenting with other hybrid systems.