Abstract: Considering it depends on so many variables, including crop genotype, environmental factors, management practices, and their interactions, predicting crop output is quite difficult. Based on environmental data and management practices, this research proposes a deep learning framework for agricultural production prediction utilizing convolution neural networks (CNNs) and recurrent neural networks (RNNs). Using historical data, the proposed CNN-RNN model was used to predict corn and soybean yield across the years 2020, 2021, and 2022. Other popular techniques included random forest (RF), deep fully connected neural networks (DFNN), and LASSO. The new model significantly outperformed all previous examined approaches, with a root-mean-square-error (RMSE) of 9% and 8% of their respective average yields. Three distinguishing characteristics of the CNN-RNN make it a potentially valuable technique for additional crop-yield prediction study.