Adoption of Machine Learning Methods for Crop Yield Prediction-based Smart Agriculture and Sustainable Growth of Crop Yield Production – Case Study in Jordan

Abstract

Crop yield prediction is significant for global food security and economic systems. Numerous algorithms for machine learning have been utilized to support crop yield prediction due to the increasing complexity of factors influencing plant growth. Machine learning (ML) models are quite tedious because the models of ML for agriculture-based are complex. This study combines several models to build a sturdy and accurate model. Linear regression predicts a measurable response using various predictors and assumes a linear relation between the response variable and predictors. This research study explores the adoption of machine learning methods for crop yield prediction and their potential to support sustainable growth of crop yields. The dataset was collected from two main sources: i) the Department of Statistics Jordan and ii) the climate change knowledge portal, which is used to train the proposed model; and the availability of large datasets has cleared the path for the application of ML techniques in crop yield prediction. Nine ML regression analysis algorithms were tested to predict the crop yield; more than one algorithm gave very good results in prediction. XGBoost, multiple linear regression, Random forest, and Lasso regression give low mean squared errors of 0.092, 0.024, 0.023, and 0.023. Crop prediction may be remarkably useful from ML algorithms, but there are many challenges. One of these challenges is the quality of the data and the data volume, where machine learning algorithms need large data. Further, because of the intricacy of agriculture systems, developing ML models can be challenging. In this research study, the strengths of optimization and machine learning are integrated to build a new predictive model for crop yield prediction. The developed integrated model in this study contributes to increasing the efficiency of crop production, and reducing prices when food shortages are found. In addition, the proposed model supports the crop prediction process, where crop prediction has a vital role in agricultural planning and procedures for making decisions. ML algorithms are an essential instrument for decision assistance for crop prediction, either in supporting decisions on the suitable to grow. The algorithm's performance may be improved by applying more innovative techniques. The developed model helps policymakers on precise forecasts, to make suitable evaluations of imports and exports to strengthen food security nationwide.