A Real-Time Precision Agriculture Framework for Soil Fertility Monitoring Using XGBoost Regression

Abstract

This work presents a new system for precision agriculture that uses machine learning and deep learning, especially for banana and mango
farms. In India, poor soil health monitoring and difficulties in managing crops effectively lower agricultural productivity. The system helps
by continuously tracking soil conditions in real time, allowing farmers to make more informed decisions. It regularly checks the health of
crops and soil, predicts soil fertility, and recommends the right amount of fertilizer to apply. The experiment results show that an XGBoost
regression model was used to estimate soil fertility, and it worked better than traditional methods like Random Forest and Decision Tree,
achieving an accuracy of 96.24%.