Final project for ECO395M Data Mining and Statistical Learning
Link to the findings: https://github.com/kimberly-hu/eco395m-data-mining-project/blob/main/eco395m_final_project.pdf
Setting the right price for an Airbnb rental is a key challenge faced every Airbnb host. In this project, I develop a predictive model to recommend listing prices to hosts, utilizing Airbnb listing data in New York City from October 2023 to March 2024, enriched with local points of interest data. I employed multiple algorithmic approaches, and identified that the XGBoost model outperformed others, achieving a minimum RMSE of approximately 67. An examination of feature importance reveals that factors such as the accommodation capacity, proximity to the city center, minimum stay requirements, and the number of bedrooms and bathrooms critically influence pricing decisions. Although the model exhibits limitations in its predictive accuracy, it provides valuable insights for hosts in setting rental prices. Several enhancements can be made to improve the performance of the model, including integrating historical booking data, applying time series analysis, and further model optimization.