<![CDATA[The rapid development of e-commerce demands a recommendation system that can help users find products that match their preferences. This study aims to implement the K-Nearest Neighbor (K-NN) algorithm in a product recommendation system to improve the personalization of e-commerce services. The K-NN algorithm works by finding similarities in behavior between users based on purchase history, then recommending products based on these similarities. The dataset used in this study consists of user transaction data, product categories, and user data, which are then processed through a cleaning and normalization stage before analysis. Testing was carried out using several variations of K values to find the optimal parameters. The experimental results showed that the value of K = 5 gave the best performance with an accuracy of 87% and an F1-score of 84.5%. In addition, the Cosine Similarity method proved effective in measuring similarities between users in sparse data. The system built is able to provide relevant recommendations with efficient computing time, showing the potential to be applied in small to medium-scale e-commerce platforms. However, the system still has limitations in handling new users (cold-start), so further development with a hybrid approach is recommended. This study shows that the K-NN algorithm is a feasible and efficient approach in user behavior-based product recommendation systems.]]>
Copyrights © 2025
