Forecasts of residential property price indices for Hong Kong through machine learning

Purpose Up to the current fall patterns that started at the end of 2021, the real estate market in Hong Kong has developed at such a rapid pace throughout the course of the previous few decades. Therefore, anticipating future property values has become a huge difficulty for both the government and investors. This is because of the current economic climate. The purpose of this study is to investigate the real estate price forecasting problem for the Hong Kong market. Design/methodology/approach Within the scope of this study, the authors use Gaussian process regressions with a range of kernels and basis functions to explore quarterly residential property price indices for Hong Kong from the fourth quarter of 1979 to the first quarter of 2024. The authors make use of cross-validation and Bayesian optimizations to train models and carry out forecasting exercises using estimated models. Findings Out of sample, the models that were developed were able to accurately predict the price indices from the second quarter of 2015 to the first quarter of 2024. The relative root mean square error for these models was 0.3080%. Originality/value It is possible that these results might be used either on their own or in conjunction with other forecasts to construct assumptions about movements in the residential real estate price indices and perform further policy research.