Performance Testing of Advanced Data Mining for Crop Yield Prediction Using SMGBR

Abstract

Crop plants are essential to economic viability, especially in agricultural areas of INDIA. Planning, resource planning, and successful practice in agriculture must possess a successful model of forecasting productive crop yield. Strategic Modified Gradient Boosting Regression (SMGBR) is proposed in this research with the objective of improving the accuracy of crop yield prediction. Unlike the typical Gradient Boosting Regression (GBR) models, the SMGBR model possesses a learning rate dynamic factor that it is better equipped to handle complex, non-linear patterns in farm-level data. This paper uses an extremely large dataset that was collected over a period of ten years (2010-2021) and contains pertinent factors like crop yield histories, advanced weather observations, and soil quality observations. Conservative performance testing was also done to check the predictability of the model for performance, efficiency, and reliability compared to conventional methods. The outcome of the work strongly indicates that SMGBR performs better than conventional methods in terms of performance but is also less computationally expensive with more dependable yield predictions. Increased reliability of the SMGBR model advantages farmers, policymakers, and farm planners the most. The model will be applied in improving better decision-making for agriculture long-term planning, resource utilization, and crop management towards economic and environmental sustainability. By the creation of sound data-driven predictions of yields, this research will improve agricultural production by limiting wastage and attaining equilibrium in ecosystems for crop development. The results are evidence of the importance of state-of-the-art machine learning methods and sound performance assessment in creating resilience and sustainability in farming communities.