Mizoram is a highly agricultural-dependent state located in the northeastern region of India, with a significant agricultural contribution. Presently, the difficulties faced by farmers are selecting crops to be grown without consideration for specific seasonal conditions, favorable climate conditions, or other factors that may affect crop output are abundantly apparent in traditional Jhum farming. Most farmers plant crops they favor without considering the elements determining the crops produced. Furthermore, this agricultural methodology is no longer profitable due to factors such as population growth, a lack of cultivable land, and adverse ecological techniques. As a result, there is an urgent need to implement cutting-edge technologies such as machine learning to improve farming operations. In this paper, we have applied a synthetic minority oversampling technique for regression with Gaussian noise (SMOGN) with various other preprocessing techniques to address these challenges. Moreover, a comparative analysis is carried out with the traditional synthetic minority over-sampling techniques for regression (SMOTER). Then, three ensemble regressor models such as the Extreme Gradient Boosting Regressor, Random Forest Regressor, and Gradient Boosting Regressors are trained and evaluated for their performance using various metrics such as MSE, MAE, RMSE, and \(R^2\) . Moreover, K-Fold cross-validation with k = 5 was employed to assess the generalizability of the models. Furthermore, the GridSearchCV optimization technique was utilized to optimize the hyperparameters of each model. Based on the experiments conducted, before and after resampling the datasets, the Gradient Boosting Regressor consistently outperformed the other models.