<p>Predicting crop yields has significant social economic and environmental ramifications while also promoting human survival. Agriculture strategy and programs aimed at ensuring food security must include crop yield prediction as a crucial element. Food security resource management coping with climate change and economic stability are just a few of the many areas it affects. In this study, an efficent model created especially for crop yield prediction. For this purpose, we used five optimization algorithms in their binary format for feature selection process known as binary dove optimization, binary particle swarm optimization, binary whale optimization, binary grey wolf optimization and binary genetic algorithm. In terms of average loss, mean fitness, optimal fitness, minimum fitness, and average deviation, the binary dove optimization algorithm produced the best results. Two datasets divided into 70:15:15 for training validation and testing were used in this study. Six distinct machine learning regression models were used to train the features that binary dove optimization chose: gradient boosting regression, kernel regressor, adaptive boosting regression, decision tree regression, random forest regression and extra trees regression. Out of all the regression models the gradient boosting model produced the best results. The dove optimization algorithm, particle swarm optimization, whale optimization algorithm, grey wolf optimization and genetic algorithm were the five optimization algorithms used to tune the gradient boosting model hyperparameters. The DOA-GB model outperformed the other models according to the experimental results with an R² of 0. 992 for the first dataset and 0. 970 for the second dataset.</p>