P‐WEV: PSO Based Weighted Ensemble Technique for Agricultural Food Price Volatility Modeling

Price volatility in agricultural commodities adds uncertainty for farmers, traders, and other stakeholders in the agricultural supply chain. Sudden price changes can disrupt income predictions, making it difficult to plan and access financing. This affects not only the farming community but also rural economies and livelihood of people involved in agriculture. The present study focused on important aspect of agricultural economic stability, that is, modeling and prediction of food price volatility. Volatility models, namely, the Generalized autoregressive conditional heteroscedastic (GARCH) model, Glosten, Jagannatan, Runkle‐GARCH (GJR‐GARCH) model, exponentially weighted moving average (EWMA) model and multiplicative error model (MEM), have been explored. In recent time, as the data becomes more complex and varied, it is difficult to fully understand price fluctuations with just one model. To tackle this, a novel approach called the Particle swarm optimization‐based weighted ensemble volatility model (P‐WEV) has been developed by combining predictions from four different models. To investigate the supremacy of the proposed model, data from 19 agricultural commodities, including cereals, pulses, oilseeds, vegetables, and spices, from various markets of India have been used. Interestingly, it has been found that the proposed model outperformed the other benchmark models. Moreover, an R package “PWEV” has also been developed for the implementation of the proposed model.