This paper explores the potential of Random Forest deep learning regression to predict the yield of major crops in the 3A agroclimatic Zone of Rajasthan, specifically in the Jaipur district. Crop yield prediction is the process of predicting yield using historical data through meteorological parameters and past yield records.Theselected region consists of a large portion of the semi-arid eastern plain, and the study focuses on six major crops i.e., barley, wheat, mustard, gram, groundnut, and moong. Time series agrometeorological data from 1991 to 2020, including rainfall, sunshine hours, temperature (minimum and maximum), and relative humidity, etc., has been collected from the Agrometeorology Observatory of Sri Karan Narendra College of Agriculture, Jobner, Jaipur. The crop yield data was obtained from the official bulletins of the Directorate of Economics and Statistics, Government of Rajasthan. The Random Forest regression, which is a supervised learning model proved to be the good performing algorithm, achieving an accuracy of 92.3%. This approach allows for optimal yield forecasting, helping farmers and policymakers better plan for crop production and management in this region. Further, attempts have been made to suggest some scientific recommendations based on the study for the benefit of farmers, policy makers and other stakeholders.

Crop Yield Forecasting, Deep Learning, Random Forest, Agroclimatic Zones