<p>Estimating crop production is crucial for food security and inevitable for agrarian policy planning for any nation. India, a country with a vast and diverse agricultural landscape, demands an efficient policy. Traditional methods of crop production estimation rely heavily on field surveys and administrative data, which are primarily conducted by unskilled workers and are time-consuming, expensive, error-prone, and may be infeasible. The objective is to propose an efficient machine learning (ML) based model for the estimation of crop production. In this endeavor, we proposed two machine learning-based models, HYBRID_reg and REG_stack, for crop production estimation. These two proposed models are constructed by fusing different traditional ML models. The performance of the proposed models is compared with other classical machine learning models, viz. Random Forest (RF), KNN Regression, Linear Regression (LR), Ridge Regression, Lasso Regression, ElasticNet, Polynomial Regression, Support Vector Regression (SVR), and Partial Least Squares Regression (PLSR). The experimental results exhibit that the performance of the proposed HYBRID_reg is not satisfactory. Meanwhile, the proposed REG_stack model shows better performance than other existing models, which can help cultivators make the right decision.</p>