Weekly price prediction of garlic and ginger using complex exponential smoothing

Abstract

Garlic and ginger crops are commercially significant spices with high export value. Over the last few years, there are more fluctuation in price, which has put growers at more risk and indirectly reduced production. Furthermore, the advanced price forecasts for both crops became a valuable input for policymakers in order to stabilize prices and a key factor of effective market intelligence service. Therefore, the study aims to identify the appropriate forecasting model for the weekly prices of garlic and ginger. For this purpose, the study used the 331 weekly observations of wholesale prices for garlic and ginger crops which were collected from the 22nd week of 2016 to the 35th week of 2022 for the Varanasi market. The statistical tests identified the presence of non-normal, nonstationary and nonlinearity features in both price series. The perusal of the literature suggested that the ARIMA and Exponential smoothing models were implied as benchmark models. Based on the presence of characteristics, the complex exponential smoothing model and Time delay neural network were selected as candidate models. From this comparative analysis, the results of accuracy measures suggested that complex exponential smoothing showed better performance compared to benchmark models and Time delay neural networks. This study recommended that complex exponential smoothing is a suitable model for forecasting the future values of garlic and ginger prices.