Crop Yield Prediction: Robust Machine Learning Approaches for Precision Agriculture

Abstract

Accurate crop yield predictions have become crucial for agriculture industries and governments to make intelligent crop management decisions and to guarantee food security. In today's data driven world, machine learning has been successfully applied in crop yield prediction, helping to determine the appropriate crops to grow in a particular area and how to maximize their production during the crop's growing and reaping seasons. Despite progress in crop yield forecasting, further research is necessary. This work represents a significant advance for the field of Precision Agriculture (PA) and has the potential to benefit farmers and agronomists greatly. By utilizing simple input parameters such as average temperature, area, average rainfall, and usage of pesticides, a helpful tool for enhancing crop yield estimation has been proposed. Machine Learning (ML) algorithms including Random Forest (RF), Extreme Gradient Boosting (XGBoost), K-Nearest Neighbor (KNN), Multilayer Perceptron (MLP), Adaptive Boosting (AdaBoost), and Decision Tree (DT) have been employed to enhance the performance of the model and offer more precise forecasting. Out of these six ML-based models, RF (MSE= 1.446, RMSE= 3.803, MAE= 2.11 and R ² =0.999) and XGBoost (MSE= 1.542, RMSE= 2.691, MAE= 2.053 and R ² =0.999) outperformed other four above-mentioned algorithms. This proposed system helps agronomist make informed decisions about their harvest, maximizing their yields and profits for tropical nations across the globe.