by Thembeka Manetje
Prof Topside Mathonsi, HoD of the Information Technology Department at the Tshwane University of Technology's Faculty of Information and Communication Technology, recently spoke at the prestigious Third IEEE Wireless Africa Conference (WAC) 2025 hosted in Pretoria. His presentation, titled "Machine Learning-Based Crop Yield Prediction: Models and Applications," addressed one of the most pressing challenges of our time—global food insecurity—and highlighted how innovative technologies like machine learning (ML) can provide solutions.
Prof Topside Mathonsi, HoD of the Information Technology Department
Prof Mathonsi’s focus on using data-driven methodologies to combat global issues is both timely and crucial. His motivation for the presentation is rooted in alarming global trends, which paint a stark picture of the future. According to the United Nations (2022), the world’s population is projected to reach 9.7 billion by 2050, driving an unprecedented demand for food.
The Food and Agriculture Organization (FAO) reports that over 280 million people across the African continent suffer from chronic hunger. This dire situation is exacerbated by several factors, including climate change, political instability and the lack of access to modern agricultural technologies.
In Africa, where the population is growing at 2.4% annually, the challenges are even more acute and the issue of food insecurity is particularly urgent, where countries like Nigeria, Ethiopia, Sudan, and Somalia frequently face severe food shortages. These shortages are compounded by unpredictable weather patterns and the devastating effects of climate change. Southern Africa has experienced devastating droughts that have severely reduced maize yields, a dietary staple for millions of people.
Meanwhile, East Africa—home to nations like Kenya, Ethiopia, and Somalia—is enduring prolonged droughts, threatening both livestock and crop yields. In South Asia, India and Pakistan are facing extreme heatwaves, negatively impacting key crops like wheat and rice. In countries like Venezuela and Haiti, ongoing economic instability and climate-related disasters have plunged many into food crises.
Prof Mathonsi’s presentation sheds light on how these environmental stressors are intensifying the pressure on farmers, many of whom lack access to modern technologies or the tools needed to make data-driven decisions to mitigate the impact. This is where machine learning comes in.
During his talk, he demonstrated how machine learning can be a game-changer in the field of agriculture. By leveraging vast amounts of historical data, weather patterns, soil quality metrics and satellite imagery, ML models can predict crop yields with remarkable accuracy. These predictions can help farmers, governments and policymakers make more informed decisions about agriculture, ultimately strengthening food security across the globe.
The applications of machine learning in agriculture are vast, offering transformative potential. With the power of accurate yield predictions, stakeholders can:
- Plan planting schedules more effectively, based on weather forecasts and soil conditions.
- Optimise resource usage, including water, fertilizers and pesticides, reducing both costs and environmental impact.
- Mitigate climate variability by identifying potential risks in advance, allowing for better preparedness and risk management.
- Strengthen food security through early warning systems and improved supply chain management, ensuring that food reaches those in need even during times of crisis.
Prof Mathonsi also highlighted the various machine learning models that can be applied to agricultural data. These include:
- Linear Regression and Decision Trees, which provide simple yet interpretable predictions.
- Random Forests and Gradient Boosting Machines (GBM), which offer more complex and accurate results from larger datasets.
- Deep Learning models, such as Recurrent Neural Networks (RNNs), that are particularly effective in handling time-series data for long-term yield forecasting.
His vision for the future of agriculture is clear: the adoption of technology-driven farming solutions is not a luxury, but an absolute necessity. He emphasised that machine learning could be a powerful ally in the fight against hunger, especially in developing countries, where access to resources may be limited, but the need for innovation is greatest.
For Africa, where agriculture remains the backbone of many economies, the potential to use technology to improve food production and distribution is immense. By embracing machine learning and other emerging technologies, African countries can strengthen their agricultural sectors, mitigate the effects of climate change and build resilient food systems that can sustain their growing populations.