An Overview of the Application of Machine Learning and Deep Learning Techniques for Agricultural Crop Yield Prediction in Terms of Methods, Data Inputs and Prospects
V. Vakdevi *
Department of Computer Science and Engineering, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
K. Satya Gayathri
Department of Computer Science and Engineering, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
V. Sowmya
Department of Artificial Intelligence and Machine Learning, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
V. Vamsi
Department of Computer Science and Engineering, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
Ch. Hari Gayathri
Department of Artificial Intelligence and Data Science, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
S. Gana Naga Bhavani
Department of Artificial Intelligence and Data Science, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
G. Sai Chandana Rani
Department of Artificial Intelligence and Data Science, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
Ch. Bhavya
Department of Artificial Intelligence and Machine Learning, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
Shaik Rangavali
Department of Computer Science and Engineering, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
S. Vyshnavi
Department of Artificial Intelligence and Data Science, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
D. Indu
Department of Artificial Intelligence and Data Science, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
M. Jahnavi
Department of Artificial Intelligence and Data Science, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
G. Bhanu Prakash
Department of Computer Science and Engineering, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
N. Sirisha
Department of Computer Science and Engineering, Lingayas Institute of Management and Technology, Madalavarigudem, Vijayawada-521212, India.
*Author to whom correspondence should be addressed.
Abstract
Correct and timely prediction of crop yields is fundamental to global food security, agricultural policy planning and the equitable management of natural resources in the phase of a rapidly changing climate. The mounting complexity of agro-environmental systems focused by soil variability, extreme weather conditions and environmental interactions has reduced dependency on traditional statistical and process based models. Over the past decade, machine learning (ML) and deep learning (DL) techniques have emerged as transformative alternatives, capable of capturing nonlinear, high-dimensional relationships across heterogeneous data sources. The data sources include remote sensing imagery, meteorological records, soil surveys and crop management accounts. This study examines the application of ML algorithms, such as Random Forest (RF), Support Vector Machine (SVM), Extreme Gradient Boosting (XGBoost) and Artificial Neural Networks (ANN) along with major DL techniques, including Convolutional Neural Networks (CNN), Long Short Term Memory networks (LSTM), hybrid CNN-LSTM models and emerging transformer based models. Key input features, such as the Normalized Difference Vegetation Index (NDVI), climatic variables, soil parameters and multi-source remote sensing data are evaluated for their influence on predictive performance. Comparisons across different crops including wheat, rice, maize and soybean reveal that ensemble and hybrid DL models consistently provide superior accuracy, with R² values commonly exceeding 0.85 in many investigations. Critical challenges including data scarcity, model interpretability deficits, geographic transferability limitations and computational demands are addressed in detail. Whereas, the role of Explainable Artificial Intelligence (XAI), transfer learning and multimodal data fusion is considered as a borderline to these limitations.
Keywords: Crop yield prediction, machine learning, deep learning, random forest, LSTM, convolutional neural network, explainable AI