AgriAI: Machine Learning Frameworks for Tailored Crop Recommendations using Soil Nutrient Parameters

Bangladesh heavily relies on agriculture, contributing 60% to its economy. Due to limited smart farming practices, the country faces challenges in achieving maximum crop productivity. To enhance productivity, nutrient profiling and crop prediction play a vital role. This study employs multiple Machine Learning (ML) techniques to recommend crops based on soil nutrient parameters. The research predicts 12 different crops using NPK values, pH, Humidity, and Temperature. State-of-the-art ML techniques including Random Forest (RF), AdaBoost (AdB), Decision Tree (DT), Support Vector Machine (SVM), K-Nearest Neighbor (K-NN), Logistic Regression (LR), XGBoost (XGB) are compared in this research. The proposed ML based frameworks are evaluated with four most popular evaluation matrices for classification such as Accuracy, Precision, recall and F1 score. Besides, hyper-parameter tuning has been carried out to enhance model performance. Findings indicated that both Random Forest and Decision Tree exhibited the superior accuracy of 99.6% and outperformed other above mentioned ML approaches.