The world’s ability to preserve food security is mostly dependent on crop production, especially India having a major producer of agricultural goods. The use of machine-learning approaches to agriculture has demonstrated enormous potential in recent years to enhance crop farming in several ways. The project intends to maximize farmers’ decision-making processes by utilizing forecasting and data-driven insights. This will empower farmers to make well-informed decisions on crop choice, irrigation, fertilization, including pest control. In this paper, many machine-learning techniques were examined for predicting agricultural yields in the state of Uttar Pradesh, India. Furthermore, it evaluates the success rate of many machine-learning methodologies, including linear regression, random forest, support vector, and CatBoost regressor, in predicting the production of 39 Indian crops from 1997 to 2019 using historical data. According to the findings, the CatBoost with 99.56% accuracy and the random forest model with 99.22% accuracy are advised for early crop yield prediction.