Airbnb Pricing in Sydney: Predictive Modelling and Explainable AI
22 Pages Posted: 25 Apr 2024
Date Written: April 24, 2024
Abstract
We employ predictive models combined with explainable AI techniques to forecast and explain Airbnb rental prices in Sydney, Australia. Our models comprise a range of algorithms from simple linear regression to advanced ensemble methods. We also consider enriching the dataset by adding new variables constructed via feature engineering. We train the models on both the original and feature-engineered datasets and compared their performance using RMSE and MAE metrics. Further, we identify sets of ‘’superior’’ forecasting models with 90 percent confidence. Ensemble methods, particularly stacking regressions, are found to outperform other algorithms using both evaluation metrics, and on both the training and test datasets. In contrast, linear models perform rather poorly, indicating the presence of more complex relationships in the dataset. Shapley values analysis reveals that factors such as property capacity, proximity to popular locations and luxury amenities lead to higher rental price predictions.
Keywords: forecasting, AirBnb rental prices, machine learning, boosting, stacking, regression
JEL Classification: C53, C50
Suggested Citation: Suggested Citation