Agricultural Product Price Prediction Based on the Quadratic Decomposition of CEEMDAN-VMD

Focusing on agricultural futures price forecasting, a prediction method based on quadratic decomposition is proposed in this paper in response to the non-stationarity, unstructured nature, and nonlinearity of agricultural price-time series data. Then drawing on the successes of deep learning in other financial domains, the quadratic decomposition of CEEMDAN-VMD that effectively addresses the non-stationarity of agricultural price-time series is introduced. And by constructing the CEEMDAN-SE-VMD-LSTM model, the paper performs an in-depth decomposition and refined processing of daily agricultural price data, successfully capturing the subtle characteristics of price fluctuations to achieve higher precision in forecasting. Moreover, the results indicates that the CEEMDAN-VMD model outperforms the comparative models in terms of forecasting accuracy for the three types of agricultural commodities.