City Recommender System: A Comparative Study of AI-Driven Approaches

This study introduces an AI-driven “Personalized City Recommender System” to recommend U.S. cities based on population, income, weather, and other factors, tailored to individual preferences. Employing logistic regression, SVM (one-versus-one and one-versus-all), Naïve Bayes, and a custom neural network, we assess their predictive performance. SVM models exhibit high accuracy, precision, recall, and F1 scores, highlighting robust predictive capabilities. Conversely, Naïve Bayes reveals limitations in accuracy and precision. The custom deep neural network consistently achieves 93.3% testing accuracy, emerging as a superior model. The custom neural network, with its high accuracy, stands out for effective urban planning. This study emphasizes the transformative role of AI in decision-making, providing individuals with data-driven insights for optimal living choices. The “Personalized City Recommender System” highlights the practical application of advanced machine learning in enhancing decision processes related to urban living ( https://github.com/riasat-mahbub/city_recommender ).