An Integrated Hedonic Pricing and Predictive Modelling Approach: Comparing ML and DL for Dublin’s Rental Market

Dublin’s persistent housing crisis has been a critical issue for a long time. While prior research has primarily focused on the property sales market, the rental market also requires attention, especially as rocketing house prices have forced many residents to rely on rented accommodation. This study bridges this gap by proposing a model integrating hedonic regression with predictive modelling to analyse rental prices in Dublin. A key challenge was overcome by conducting web scraping to construct a detailed dataset encompassing core features summarised from the literature review. The dataset was further enhanced through spatial analysis by integrating proximity measures to key external amenities. The findings in this study revealed a significant rental burden issue within Dublin’s housing market, with the Rent-To-Income Ratio (RTR) across all household types exceeding the affordability threshold of 30%. Key features that influence rental prices were identified, including property types, building energy ratings (BER), number of bedrooms and bathrooms, and accessibility to neighbourhood and location facilities and amenities. Among Linear Regression, Decision Tree, Random Forest, SVR, XGBoost, LightGBM, GBR, and RNNs models, Light-GBM was found to achieve the optimised predictive accuracy, with an R² of 0.79, MSE of 106189.96, MAE of 234.9, and RMSE of 325.97.