Predictive Modeling of Off-Campus Student Housing Rentals Using Machine Learning

Finding affordable and suitable housing poses a significant challenge for students, particularly those studying abroad or away from their hometowns. Decision-making is influenced by various factors, including location, room size, and available amenities. While extensive research has explored price predictions for different accommodation types, studies specifically targeting rental predictions for student housing remain limited and often region-specific. This study aims to develop a machine learning model that predicts price ranges for private accommodations tailored to college students. The model leverages a diverse set of features, including personal demographics, general accommodation details, and renter satisfaction with various facilities, enhancing its generalizability across different locales. Traditional learning algorithms were used for this multi-class classification problem, and data resampling methods were utilized to tackle the imbalanced class distribution and improve model accuracy. Results show that logistic regression had superior performance compared to the other algorithms, and the use of a hybrid resampling technique significantly improved recall rates.