Evaluation of agricultural sustainability in Maharashtra using machine learning techniques

Assessing agricultural sustainability is essential for advancing the Sustainable Development Goals; integrating innovative agricultural strategies with SDG targets can enhance rural livelihoods, mitigate climate change impacts, and achieve inclusive and equitable development globally. This study evaluates sustainability across all 35 districts in Maharashtra by employing the Sustainable Livelihood Security Index framework. The main objective is to quantify district-level variations in sustainability by using SLSI, which integrates ecological, economic, and social factors to assess agricultural sustainability across different periods. Supervised machine learning techniques, including Classification and Regression Tree and Random Forest Model, were employed to predict SLSI values. RFM achieved high accuracy (R2 = 0.9842), with the EEI having the most significant impact on SLSI predictions. The results indicate that districts like Pune and Kolhapur consistently performed well, while Satara and Ahmadnagar exhibited fluctuating trends in SLSI. Gadchiroli, Sindhudurg, and Washim districts continuously performed poorly and remained at the bottom. Notable improvements were observed in districts such as Aurangabad and Nashik, while others, like Bhandara and Wardha, experienced declines. The overall SLSI for Maharashtra showed a marginal increase from 0.616 in 2010 to 0.630 in 2023, reflecting limited progress in agricultural sustainability across the state. The findings emphasize the need for targeted ecological restoration and socio-economic development interventions to strengthen sustainability, particularly in underperforming districts.