Research on Ginger Price Prediction Based on Prophet- Support Vector Machine (SVM) Combination Model

Abstract

The traditional time series forecasting algorithm is difficult to solve the nonlinear price forecasting problem, and the modern time series forecasting algorithm is very dependent on data. In this study, we propose a price forecasting model based on Prophet SVM and take ginger as an example. In the process of model construction, the price series is decomposed into trend term, periodic term, and random term by using the Prophet model, and then the prediction of trend and the periodic term is carried out by Prophet, and the random term is predicted by SVM. Finally, the forecast value of each component is combined to get the final prediction result. The experimental results show that the root mean square error (RMSE is 0.39) of the Prophet SVM combination model is less than that of Prophet (RMSE is 0.45) and that of SVM (RMSE is 1.47). The combined model has higher prediction accuracy and better prediction performance. Prophet SVM has higher accuracy in agricultural product price prediction, which provides an important scientific basis and theoretical support for agricultural product price prediction.