<![CDATA[Information overload in tourism applications poses significant challenges for users selecting relevant destinations from numerous options. This research implements Collaborative Filtering (CF) to address information overload in the Campsite Indonesia mobile application, where users face difficulties choosing from 246 camping locations. Three CF variants are evaluated: User-Based CF, Item-Based CF, and Hybrid Collaborative Filtering. The dataset comprises 746 users, 246 camping locations, 350 explicit feedback interactions (likes), and 7,306 implicit feedback interactions (views) from August 2022 to July 2025, with 94.05% sparsity in the user-item interaction matrix. The research employs CRISP-DM methodology encompassing data preparation, modeling, evaluation, and deployment phases. Experimental results demonstrate that Item-Based CF achieves superior performance with Hit Rate@10 of 0.2222 and NDCG@10 of 0.0743, significantly outperforming User-Based CF (HR@10: 0.0556, NDCG@10: 0.0215) and Hybrid CF (HR@10: 0.0000, NDCG@10: 0.0000). Item-Based CF also exhibits the highest coverage (41.10%) with 60 unique recommended locations. The system is deployed through a Flask-based REST API server with five endpoints for recommendation scenarios. This research contributes domain-specific insights for camping location recommendations in developing countries. ]]>
Copyrights © 2025
