A KNN-Based Stacking for Tourism Destination Recommender System Based on User Demographic Data

The tourism destination recommendation system plays a crucial role in helping tourists find destinations that match their preferences.However, existing systems struggle with managing complex demographic data, leading to suboptimal prediction accuracy and reduced user satisfaction.To address this challenge, this study introduces a novel ensemble learning approach called KNN-Based Stacking, which enhances recommendation accuracy by effectively capturing diverse demographic patterns.Unlike conventional methods that rely on a single predictive model, the proposed approach integrates multiple KNN (k-Nearest Neighbors) variations with different neighborhood sizes ( = 10, = 15, and = 20) as base learners and optimizes prediction results using Logistic Regression as a metalearner.This strategy allows the model to leverage the strengths of different KNN configurations, improving its ability to generalize across varied user profiles.Experiments were conducted on a dataset of 1,000 users with demographic attributes linked to 14 tourism destinations.The performance of the proposed KNN-Based Stacking model was compared against Nave Bayes, CART (Classification and Regression Tree), SVM (Support Vector Machine), and Random Forest, demonstrating superior performance across all metrics.The model achieves 86.89% accuracy, 87.18% precision, 86.89% recall, and 86.93% F1-score, confirming its effectiveness in improving personalized tourism recommendations.This study contributes a novel ensemble learning strategy tailored for tourism recommendation systems, offering a more adaptive and accurate approach for handling complex user demographics.Future research can explore deep learning-based hybrid methods and real-time optimization techniques to further enhance recommendation performance.