Forecasting maize yield using ARIMA-Genetic Algorithm approach

Abstract

Maize is widely cultivated throughout the world and has highest production among all the cereals. India is the sixth largest producer of maize in the world, contributing 2% of global production and accounting for 9% of the total food grain production in the country. Based on increasing growth rates of poultry, livestock, fish, and milling industries, the demand for maize is expected to increase from the current level of 17 to 45 million tons by 2030. To understand the growing pattern and economics of crop production, it is necessary to predict crop yield using statistical models and geographic information system soil mapping and the impacts of insect and pest damage. In this study, the focus was to forecast maize yield in India using an autoregressive integrated moving average (ARIMA) model and genetic algorithm (GA) approach. GA simulates the evolution of living organisms, where the fittest individual dominates the weaker ones by mimicking the biological mechanism of evolution, such as selection, crossover, and mutation. GA has successfully been applied to solve optimization problems. The study reveals that implementation of GA in ARIMA enhances the prediction accuracy of the model.