Predicting Soil Moisture Using Gradient Boosting with Linear Base Learner and Tree-Based Method

Soil moisture prediction plays a crucial role in agriculture and environmental monitoring by optimizing irrigation improving crop yield, and preserving water resources. The machine learning model is used to predict soil moisture. This work presents a hybrid, that combines Gradient Boosting with a linear base learner for soil moisture prediction. By including meteorological parameters such as temperature, humidity, and UV index, the model provides better predictive performance compared to the traditional Gradient Boosting Regressor, which also gives a higher accuracy and lower error rates. The hybrid model also improves generalization through balancing bias and variance, thus being more robust and less prone to overfitting. Hybrid models can identify complex patterns in data, but an individual model can overfit due to its complexity or because of poor regularization. The study emphasizes the role of hybrid models in potentially advancing agricultural and environmental decision-making by showing the practical applicability of simulated and predicted soil moisture values.