<![CDATA[Recommendation systems play a crucial role in helping users choose complex and diverse products, such as laptops, which have numerous and varied technical attributes. This research aims to implement a Hybrid-Based Recommendation method that combines Content-Based Filtering (CBF) and Collaborative Filtering (CF). CBF is implemented using TF-IDF Vectorization and cosine similarity to recommend laptops based on technical attribute similarity. Concurrently, CF uses Singular Value Decomposition (SVD) to predict user preferences based on rating history. A Cascade Hybrid strategy is applied by filtering initial candidates from CBF and then re-ranking them using rating predictions from CF. The dataset comprises laptop data and user ratings obtained from Kaggle. Evaluation is performed using the NDCG metric to measure the relevance order of recommendations and MAPE to assess prediction accuracy. The research results indicate that this hybrid system is capable of generating relevant and personalized recommendations, with an NDCG value of 0.9838 and a MAPE value of 27.94%. The study concludes that the integration of CBF and CF through a hybrid approach effectively produces relevant and effective recommendations. For future development, exploring other hybrid methods, parameter optimization, and direct user testing are suggested.]]>
Copyrights © 2025
