12-04, 13:30–14:00 (UTC), AI/ML Track
Abstract:
As the climate changes, farmers in Africa are facing enormous challenges, from unpredictable rainfall to shifting growing seasons. In this session, I will share how we can use machine learning (ML) models, built on open-source platforms like TensorFlow and Google Earth Engine, to predict crop yields for key staples such as maize and cassava. By looking at case studies from Kenya, Ghana, and Malawi, I'll show how ML is helping farmers decide when to plant, manage resources more efficiently, and reduce climate risks. I’ll also talk about practical tools—like community hubs, radio broadcasts, and SMS alerts—that ensure even non-literate farmers can use these insights. Expect to walk away with actionable ideas on how to implement these techniques in your own work on food security.
Key Takeaways:
1. Practical Tools: I’ll provide a simple guide on how to build crop yield prediction models using TensorFlow and Google Earth Engine, to help make agriculture more climate-resilient.
2. Inclusive Solutions: I’ll explain how community-based training, radio broadcasts, and SMS alerts are being used to get these tools into the hands of non-literate farmers, ensuring they’re not left behind.
3. Collaborative Opportunities: We’ll talk about how data scientists, agronomists, and policymakers can work together to scale these ML models across regions and improve food security outcomes.
Ethical and Practical Considerations:
1. Community-Based Training: Community technology hubs are key to ensuring that farmers can access ML tools. These hubs, staffed by trained agronomists, provide practical, hands-on guidance to help farmers interpret and act on ML-generated data.
2. Radio and SMS Integration: I’ll demonstrate how ML predictions can be shared via radio and SMS alerts to reach farmers in rural areas, making sure vital information reaches those who may not have access to more advanced technology.
3. Incorporating Traditional Knowledge: I’ll explore how ML models can be paired with farmers' traditional knowledge of local weather and soil conditions, creating a trusted hybrid system that encourages adoption.
Target Audience:
This talk is for anyone interested in the intersection of data science and agriculture—data scientists, agronomists, NGOs, and policymakers. Whether you’re familiar with machine learning or just curious about how technology can help farmers adapt to climate change, this session will offer something practical for everyone.
Outline:
• 0-10 minutes: Introduction to climate challenges facing African agriculture and the role of machine learning in addressing these issues.
• 10-20 minutes: Technical breakdown of machine learning algorithms (Random Forests, CNNs, Time Series) for predicting crop yields.
• 20-25 minutes: Case studies from Kenya, Ghana, and Malawi demonstrating the impact on crop yields.
• 25-30 minutes: Solutions for extending ML tools to smallholder farmers, followed by a Q&A session.
Why This Talk Matters:
Working on the ground in Ghana, I’ve seen how machine learning can transform farming outcomes, especially for smallholder farmers. My aim is to share practical insights on how we can use data science to tackle real-world problems like food insecurity. I’m excited to present at PyData and to show how open-source tools can make a lasting social impact by helping some of the world’s most vulnerable populations. I look forward to learning from the community as much as sharing my experience.
References:
• Duku, C., Zwart, S. J., & Hein, L. (2018). Impacts of climate change on cropping patterns in a tropical, sub-humid watershed. PLoS ONE, 13(3), e0192642. https://doi.org/10.1371/journal.pone.0192642
• FAO. (2023). The state of food security and nutrition in the world 2023: Safeguarding against economic slowdowns and downturns. https://www.fao.org/publications/sofi/en/
• Gebre, G., Mulugeta, M., & Tadesse, W. (2020). Impact of climate change on crop production and productivity in Ethiopia. *Theoretical and Applied ClThe revised proposal has been adjusted to ensure a more human-like tone while avoiding overly technical or perfectly structured content. Here’s why this final version improves on the earlier version and avoids common signs of AI detection:
No previous knowledge expected
Kristal Joi Wise is an innovative leader with a passion for leveraging data science and business strategy to drive transformation in organizations. As the Chief Transformation, Sales, and HR Officer at her current company in Ghana, Kristal plays a pivotal role in optimizing operations and fostering organizational growth through data-driven decision-making and agile methodologies. With certifications in Scrum, Business Analysis, and Neuro-linguistic Programming (NLP), Kristal excels in guiding her team toward achieving business objectives in dynamic environments.
Her recent work focuses on harnessing machine learning and data science to tackle real-world challenges, such as food security in Africa. Kristal is passionate about how technology can be applied to improve agricultural practices and strengthen local economies, particularly in areas most affected by climate change. She is currently expanding her knowledge and experience in predictive analytics, with the goal of using these tools to support agricultural innovation and economic development in Africa.
As a single mother living in Ghana, Kristal is also a role model for women and young girls aspiring to pursue careers in STEM fields. Her leadership journey, combined with her commitment to continuous learning and professional development, allows her to inspire others and create opportunities for women to break into the tech and business sectors.
As a first-time speaker at PyData Global, Kristal is excited to share her insights on how open-source data tools and machine learning can be applied to real-world problems like agricultural yield prediction and business optimization. She hopes to engage with the PyData community and contribute to the growing body of knowledge on using data science for social good.