AfriCDSA Hackathon 2025

Problem Statement

Cocoa is a key economic driver for millions of smallholder farmers in Africa, particularly in countries like Ghana and Côte d'Ivoire, which together account for over 60% of global cocoa production. Cocoa Swollen Shoot Virus Disease (CSSVD) and other cocoa diseases threaten not only individual livelihoods but also the broader agricultural economy.

Challenge Objectives

• Develop machine learning models to accurately predict all diseases present in images of cocoa
• Create models that can generalize well, even when encountering new diseases not seen in the training set
• Ensure models operate efficiently on edge devices such as entry-level smartphones

Impact

By developing machine learning models capable of accurately identifying multiple cocoa diseases, this challenge will empower farmers with real-time, AI-driven diagnostics accessible through their mobile phones. The ability to detect new and emerging diseases beyond those seen in the training set will improve early intervention strategies, minimizing crop losses and reducing reliance on reactive pesticide use.

Problem Statement

In the remote Kalam region of Pakistan, micro-hydropower plants (MHPs) are the backbone of energy generation for rural communities. But nature doesn't always play along. Water flow fluctuates, maintenance disrupts supply, and the weather throws in its own surprises.

Challenge Objectives

• Build a model that accurately predicts energy load generation (in kWh)
• Utilize MHP data including voltage, current, power factors, and energy metrics
• Incorporate climate indicators like temperature, dew point, wind speed, and precipitation

Impact

Predicting energy needs means less waste, fewer blackouts, and smarter resource management—all crucial for keeping the lights on in off-grid communities. Understanding how climate affects energy demand, and what that means for sustainable power planning, is a critical insight for sustainable power projects all over the world.

Problem Statement

Data for this challenge has been collected over 11 months. If a user is the first user to purchase at a merchant, the app asks the user to manually classify the merchant. The next user to purchase at that merchant has the opportunity to confirm the suggestion or enter a new categorization.

Challenge Objectives

• Create a machine learning algorithm that classifies each purchase into one of 13 different categories
• Work with verified transaction data from MPESA SMS receipts
• Handle both verified and unverified transactions effectively

Dataset Information

• ~400 purchases in training data
• ~600 purchases in test data
• ~10,000 unverified transactions
• All transactions have been verified by Alvin as correctly classified

Problem Statement

In Third world countries, climate-sensitive waterborne diseases such as typhoid, amoebiasis, diarrhoea, schistosomiasis, and intestinal worms pose significant health risks, particularly for vulnerable populations like women and children. These diseases are exacerbated by poor water quality, inadequate sanitation, and changing climate patterns.

Challenge Objectives

• Develop a machine learning model to predict outbreaks of climate-sensitive waterborne diseases
• Utilize datasets comprising water sources, toilet quality, waste management, health facility data, and climate information from 2019 to 2023
• Focus on a specific region in Tanzania

Impact

This predictive capability will enable governments and health organisations to implement timely, targeted interventions and optimize resource allocation, ultimately reducing disease incidence and enhancing public health resilience.