AI-Driven Crop Yield Prediction: Optimizing Agricultural Practices using Machine Learning Models

Machine learning is the solution to effectively predict crop yields and make informed choices concerning the cultivation and management of crops by farmers. Due to the complex nature of farming variables, it requires much for analysis, but machine learning could facilitate data-driven decision-making. However, the primary aim is to create an accurate prediction-making model in machine learning. Random forests, decision trees, linear regression, and XGBoost were different machine learning approaches we tried using on historical data from different geographical sites to find out how best to predict crop yields. Our goal was to build a robust predictively accurate model, allowing producers to make data-driven decisions about their resource management, as driven by, say, approaches such as yield estimation from soil properties. The individual models' performance was further evaluated by R-squared (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) for training and independent test datasets. The XGBoost model had a significantly higher R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score than the other methods, according to our findings. The accuracy of the XGBoost model was subsequently increased by further fine-tuning of parameters. Finally, the model has an innovative Flask-based web interface for easy interaction. This interface allows users to readily input pertinent data and obtain reliable crop yield predictions. This research demonstrates that an optimised XGBoost-based approach can significantly improve yield prediction, thereby contributing to the improvement of economic outcomes and the increase in agricultural productivity.