Discrete interaction grey forecasting model with dual perspective coupled seasonal dummy variables and its application

Purpose To address challenges in forecasting seasonal time series, especially in agricultural drought forecasting, this paper proposes the discrete interaction grey forecasting model with dual perspective coupled seasonal dummy variables (DS-DVGM(1,1,T)). Design/methodology/approach The DS-DVGM(1,1,T) model is built upon the theoretical framework of the discrete grey forecasting model, incorporating a dual processing mechanism to account for the dual interaction effect and nonlinear characteristics of the system. Seasonal dummy variables are introduced to portray the fluctuating tendency of seasonal series. The parameter estimation method for the new model is discussed, and the Differential Evolutionary algorithm is used to solve the linear and hyper-parameters of the model. Findings The DS-DVGM(1,1,T) model outperforms other forecasting models, accurately describing the seasonal pattern of soil moisture in six cities of Henan Province, with forecast results closely matching actual drought conditions. Practical implications The DS-DVGM(1,1,T) model provides a reliable forecasting benchmark for agricultural drought monitoring. By accurately forecasting drought conditions, farmers and policymakers can take timely and appropriate actions to mitigate the impacts of drought on agriculture and grain security. Originality/value The DS-DVGM(1,1,T) model offers a new perspective and methodology, overcoming the limitations of existing models in capturing seasonal trends and nonlinear features, aiding agricultural drought monitoring and management.