Empirical Analysis of Crop Yield Prediction Using Hybrid Model

Over the past few years, the agricultural sector has undergone a remarkable transformation through the incorporation of state-of-the-art technologies. In the year 2023, the emergence of machine learning (ML), robotics, artificial intelligence (AI), and drones is poised to redefine the future of agriculture, propelling us into an era characterized by smart farming. Smart farming is an emerging concept that leverages cutting-edge information technologies to enhance the efficiency and effectiveness of agriculture. But predicting crop yields poses a formidable challenge within the realm of agriculture. It holds a key role in decision-making globally. Crop yield prediction relies on factors encompassing soil quality, environmental factors, crop-specific parameters, and meteorological conditions. In recent studies, machine learning techniques have been employed to forecast crop yield, encompassing techniques like decision trees, multivariate regression, association rule mining, and artificial neural networks. In this paper, an empirical research of crop yield prediction is made by leveraging machine learning (ML) techniques and ensemble-based methods. We explore the potential of various ML techniques comprising K-nearest neighbor (KNN), decision tree regressor (DTR), and support vector regressor (SVR) for prediction, and the final prediction is made using ensemble algorithms like AdaBoost, gradient boosting (GB), and extreme gradient boosting (XGB), for accurate and robust crop yield forecasting. The result shows that the XGBoost algorithm outperforms gradient boosting and AdaBoost ensemble algorithms.