In the heart of India’s agricultural landscape, a technological revolution is brewing, promising to reshape how farmers and policymakers approach crop yield prediction. At the forefront of this innovation is Anuradha Yenkikar, a researcher affiliated with the School of Engineering at Amity University Dubai Campus and the Department of Computer Science and Engineering (AI) at Vishwakrma Institute of Technology in Pune. Her latest work, published in MethodsX, introduces an explainable AI-based hybrid machine learning model designed to enhance crop yield predictions while ensuring transparency and interpretability.
Agriculture is the backbone of India’s economy, contributing significantly to the GDP and employing a vast population. Crops like rice are pivotal for food security, making accurate yield predictions crucial for sustainability. Traditional machine learning models have improved forecast accuracy, but their lack of interpretability has been a barrier to widespread adoption. Yenkikar’s research addresses this challenge by integrating Explainable AI (XAI) techniques into a hybrid model that combines Random Forest (RF), Long Short-Term Memory (LSTM), and XGBoost algorithms.
The hybrid model leverages SHAP (SHapley Additive Explanations), LIME (Local Interpretable Model-Agnostic Explanations), and Counterfactual Analysis to provide clear, actionable insights. “The goal is to make AI-driven predictions more transparent,” Yenkikar explains. “By doing so, we can build trust and facilitate better decision-making among farmers and policymakers.”
The model was tested on a large-scale, multi-year agricultural dataset comprising over 246,000 records across 33 states. This dataset, provided by the Indian Agriculture Department, includes a wealth of information on various crops, seasons, and climatic factors. The results were impressive, with the model achieving high accuracy (R² = 0.9827 for crop yield and 0.9721 for rice yield), outperforming existing models.
One of the standout features of Yenkikar’s research is the development of the ‘E-Kisan’ web interface. This platform delivers actionable insights directly to farmers and policymakers, making complex data accessible and understandable. “The E-Kisan interface is designed to be user-friendly,” Yenkikar notes. “It translates technical jargon into practical advice, helping farmers make informed decisions.”
The implications of this research are far-reaching. For the energy sector, accurate crop yield predictions can lead to more efficient resource allocation and reduced waste. Farmers can optimize the use of fertilizers, water, and other inputs, leading to cost savings and environmental benefits. Policymakers can better plan for food security and economic stability, ensuring that the agricultural sector remains a robust contributor to the nation’s economy.
As we look to the future, Yenkikar’s work paves the way for more transparent and effective AI applications in agriculture. The integration of XAI techniques into machine learning models can revolutionize how we approach crop yield prediction, making the technology more accessible and trustworthy. This research, published in MethodsX, which translates to MethodsX, sets a new standard for agricultural technology, promising a more sustainable and prosperous future for farmers and the broader economy.