Recommendation System Based Collaborative Filtering for Deciding Travelling Place

This Indonesia has a vast number of tourist destinations, and it can be overwhelming for people to decide where to go, either because they have many options to choose from or because they are bored with places they have visited before. To address this issue, this research uses collaborative filtering-based recommendation systems to provide users with personalized recommendations based on their previous ratings. The research will compare five recommendation system algorithms: the Hybrid Algorithm with ContentKNN (Content K Nearest Neighbors) and RBM (Restricted Boltzmann Machine), SVD++ (Singular Value Decomposition++), Efficient Deep Learning, Scalable Deep Learning, and Naïve Bayes. These algorithms will be tested based on their evaluation metrics, namely RMSE (Root Mean Squared Error) and MAE (Mean Absolute Error). The model that has the best evaluation metrics will be selected to provide travelling place recommendations. To develop algorithm models for the recommendation system, the research will use the CRISP-DM (Cross-Industry Standard Process for Data Mining) model. Based on the workflow, Efficient Deep Learning has the best RMSE and MAE value from all of the models, with these following parameters: 250 neurons and SELU (Scaled Exponential Linear Unit) activation type. Future works for this research includes additional rating dataset, adding different combinations of fields, distance metrics, and other evaluation metrics.