<![CDATA[The rapid development of internet technology has driven increased adoption of e-commerce, yet companies face challenges in enhancing users' shopping experiences. To assist users in finding products that match their preferences, relevant recommendation analysis is crucial. This research compares the effectiveness of K-Nearest Neighbors (KNN) and Singular Value Decomposition Plus Plus (SVD++) algorithms for e-commerce product recommendations using the Tokopedia Product Reviews dataset from Kaggle, which contains 40,893 reviews. The study includes data collection and preprocessing steps such as removing duplicates, replacing missing values with the average, and normalizing ratings. KNN and SVD++ are then applied to predict ratings using cosine similarity and factor matrices. Evaluation using Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) shows that SVD++ outperforms KNN, achieving a lower MAE of 0.161176 and RMSE of 0.185252, compared to KNN's MAE of 0.163964 and RMSE of 0.197045. This indicates that SVD++ is more effective in delivering accuracy and capturing data complexity. The findings highlight the potential to enhance recommendation effectiveness in e-commerce, improving user satisfaction by efficiently matching products to preferences.]]>
Copyrights © 2024
