SLICKMAN-TYRUS/south-sudan-conflict-food-security-ml

Predicting food security phases in South Sudan using Machine Learning and deep learning. # Predicting Food Insecurity Risk from Conflict Dynamics in South Sudan *A comparative machine learning and deep learning approach for humanitarian early-warning systems.* --- # Project Overview Food insecurity remains one of the most pressing humanitarian challenges in South Sudan, where prolonged armed conflict continues to disrupt livelihoods, agricultural production, and access to essential resources. Humanitarian organizations rely on periodic food security assessments to guide emergency interventions; however, these assessments often provide limited lead time for proactive response. This project investigates whether **historical conflict event data can be used to predict county-level food security crises one month in advance**. Using conflict records from the **Armed Conflict Location & Event Data (ACLED)** project and food security classifications from **FEWS NET / IPC**, a complete machine learning pipeline was developed to compare traditional machine learning models with deep learning approaches for humanitarian forecasting. Rather than replacing existing humanitarian assessments, the project demonstrates how conflict data can serve as an additional early-warning signal to support more timely, evidence-based decision-making. --- # Research Question > **Can county-level conflict event data predict the likelihood of a food security crisis one month in advance?** --- # Mission Alignment This project supports the broader goal of improving humanitarian decision-making and promoting sustainable development in conflict-affected regions. By identifying patterns that precede food insecurity, the developed models demonstrate how machine learning can assist governments, humanitarian organizations, and peacebuilding agencies in prioritizing interventions before conditions deteriorate. --- # Project Objectives The project was designed to: * Develop an end-to-end machine learning pipeline for conflict-driven food security forecasting. * Compare the perfor …