Machine Learning Models for Recommending Crops to Farmers in the Face of Unbalanced Data

In machine learning, oversampling is a technique used to balance out unbalanced datasets. This technique reduces bias and improves model accuracy by making duplicates of the minority class occurrences. The discipline of crop analysis and prediction is expanding quickly and is essential to improving agricultural operations. Crop suggestion is essential to agriculture because it gives farmers the knowledge and ability to choose the best crops for their particular climate and area. This research presents a machine learning model-based approach for crop recommendation based on many parameters. Farmers can no longer select the most suitable crop based on the properties of the soil and other factors. In this work, we suggest using machine learning techniques after pretreatment with the Synthetic Minor Oversampling technique (SMOTE). By assessing the ROC receiver’s operational characteristic, the method put forth here increases learning accuracy and enables improved test outcomes. This work focuses also on accuracy, precision, recall, and F1-Score to analyze the machine learning model.