Optimizing Agriculture Crop Yield Prediction and Classification Using Machine Learning and Deep Learning

Efficient prediction of crop yield is crucial for the sustainability of agricultural practices. It plays a crucial role in ensuring food security and aids in maximizing resource management efficiency. This work introduces two machine learning (ML) models, one for improving the accuracy of yield prediction and another one for crop Classification, and one Deep Learning model. These models integrate a comprehensive range of factors, encompassing crop-specific attributes, prevailing weather conditions, geographical coordinates, and soil properties, for yield prediction. In this work a Global Dataset was used. The data set has details of 10 crops from around 100 countries from the year 1990 to 2013. For yield prediction, XGBoost Regressor ML model has outperformed other ML models with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{R}^{2}$</tex> score of 0.99879 and MAE of 0.00672 with Percentage MAE of 1.3324% for the whole data, with a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{R}^{2}$</tex> score of 0.890 (for Yam crop) to 0.973 (for Sweet Potato crop) on crop-wise data. While DL model, Deep Neural Network (DNN) scored <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{R}^{2}$</tex> score of 0.9555 and MAE of 0.0413 with Percentage MAE of 8.27%. For Crop Classification, Decision Tree Classifier scored an accuracy of 91.11%. Through data-driven decision-making, this work provides a solid answer for precision agriculture, enhancing yield prediction and promoting sustainable farming methods.