Predicting the Producer Price Index (PPI) in China Using Gaussian Process Regression Frameworks Parameterized through Bayesian Inference

Forecasting China’s producer price index (PPI) offers early insight into inflationary trends and cost pressures that affect both domestic economic stability and global supply chains. Accurate PPI predictions enable policymakers, investors and firms to make more informed decisions on monetary policy, pricing strategies and resource allocation. This study introduces a novel predictive scheme based on Gaussian process regression (GPR) wherein model hyperparameters are inferred through a Bayesian framework, thereby enabling the forecasting mechanism to dynamically adapt to latent market volatilities and unobserved structural changes. By accounting for these evolving dynamics, the proposed approach more accurately represents shifting patterns in China’s PPI. Empirically, the analysis draws on a comprehensive monthly dataset covering the period from October 1996 to February 2025, thus incorporating multiple phases of regulatory reform, industrial restructuring and macroeconomic transformation. For validation, an out-of-sample testing interval from June 2019 through February 2025 is employed, producing a relative root mean square error (RRMSE) of 0.7040%, a root mean square error (RMSE) of 0.7094, a mean absolute error (MAE) of 0.5446 and a correlation coefficient (CC) of 0.98903. To the best of our knowledge, this represents the first application of a Bayesian-inference-parameterized GPR model to forecast China’s PPI. Beyond advancing theoretical discourse on machine-learning-based price prediction, the methodology offers a flexible analytical template that can be adapted to analogous macroeconomic time-series forecasting tasks.