Advanced Soil Moisture Predictive Methodology in the Maize Cultivation Region using Hybrid Machine Learning Algorithms

Abstract

The moisture level in the soil in which maize is grown is crucial to the plant's health and production. And over 60% of India's maize cultivation comes from the states of South India. Therefore, forecasting the soil moisture of maize will emerge as a crucial factor for regulating the cultivation of maize crops with optimal irrigation. In light of this, this research provides a unique Improved Hybridized Machine Learning (IHML) model, which combines and optimizes several ML models (base learners-BL). The convergence rate of all the considered BL approaches and the preciseness of the proposed approach significantly enhances the process of determining the appropriate parameters to attain the desirable outcome. Consequently, IHML contributes to an improvement in the accuracy of the overall forecast. This research collects data from districts in South India that are primarily committed to maize agriculture to develop a model. The correlation evaluations served as the basis for the model's framework and the parameter selection. This research compares the outcomes of BL models to the IHML model in depth to ensure the model's accuracy. Results reveal that the IHML performs exceptionally well in forecasting soil moisture, comprising Correlation Coefficient (R2) of 0.9762, Root Mean Square Error (RMSE) of 0.293, and Mean Absolute Error (MAE) of 0.731 at a depth of 10 cm. Conceptual IHML models could be used to make smart farming and precise irrigation much better.