CONTEXT: Recognizing the important role of technology in modern farming practices, we aimed to create a solution that could assist farmers in making informed decisions about crop selection and cultivation methods.

OBJECTIVE: The primary objective of our project is to design and implement a robust machine learning model capable of accurately predicting crop yields based on various environmental and agricultural factors. It leverages historical data and advanced predictive analytics techniques like the mineral content of the soil along with the parameters N-P-K, pH, temperature, rainfall, and humidity.

MATERIALS AND METHODS: To achieve this objective, we gathered a diverse dataset comprising information on weather patterns, soil characteristics, crop types, and agricultural practices from different regions. We then implemented various machine learning algorithms, including DT, RF, and SVM, LDA, LR, GNB, SVC, and KNN, to train and evaluate the predictive models. To improve the standalone model's performance, we have proposed one ensemble learning technique. We also used the ensemble learning algorithm to increase the performance of the model.

RESULTS: In this paper, we have used several machine learning classifiers for crop recommendation, and these are DT, GNB, SVM, LR, and RF. We also measured their performances (accuracy, Precision, F1-Score, and Recall) and found their accuracy (90%, 99%, 97%, 95%, 98%).

CONCLUSION :In this article, GNB's accuracy outperforms, i.e., by 99%, the other mentioned classifiers in recommending crops.