Forecasts of composite real estate price indices through gaussian process regressions

The Chinese real estate market has grown so rapidly in the last 10 years that forecasting future property values has become a major issue for both the government and investors. In this study, we use Gaussian process regressions with a range of kernels and basis functions to explore monthly composite real estate price index projections for 10 major Chinese cities from 7M2005 to 4M2021. We employ cross-validation and Bayesian optimizations to do forecast exercises. The developed models accurately forecast the 10 price indices from 5M2019 to 4M2021 out-of-sample, with relative root mean square errors spanning from 0.0203% to 0.2908%. Our results might be applied separately or in conjunction with other forecasts to develop hypotheses regarding patterns in the composite real estate price index and conduct further policy research.