arikos/uganda-road-crash-intelligence

# Uganda Road Crash Risk & Severity Intelligence Platform 🚨 ## About The Project The **Uganda Road Crash Risk & Severity Intelligence Platform** is a decision-support system designed to transform road safety management in Uganda from reactive statistical reporting to proactive, data-driven prevention. Powered by a class-weighted **LightGBM** machine learning model and deployed via **Streamlit**, the system evaluates crash dynamics—such as corridor location, vehicle class, contributory cause, time of day, and environmental conditions—to predict crash severity levels (*Minor*, *Serious*, or *Fatal*) and pinpoint high-risk hazard windows across the national road network. --- ## Key Features * **🎛️ Single Scenario Risk Evaluator:** Allows traffic officers and safety analysts to input individual crash parameters and receive instant predicted severity probabilities alongside active risk driver warnings. * **🗺️ Spatial-Temporal Network Heatmap:** Systematically simulates risk across all 7 major Ugandan highway corridors and 24 hours of the day (168 matrix nodes) to map exactly *where and when* severe crashes are most likely to occur. * **📁 Bulk CSV Batch Profiler:** Enables organizations to upload unlabelled datasets containing thousands of incident logs or trip records, score them automatically, and export an enriched CSV report with predicted severity labels and individual fatality confidence scores. --- ## Real-World Impact By moving beyond historical reporting, this platform delivers actionable intelligence to key stakeholders: * **Uganda Police Force (UPF):** Optimizes patrol positioning, speed trap locations, and breathalyzer checkpoints during peak risk hours. * **Ministry of Health & Emergency Medical Services (EMS):** Pre-positions ambulances along high-fatality corridors to reduce trauma response times during the critical "golden hour." * **UNRA & Ministry of Works and Transport:** Identifies high-risk blackspots to prioritize civil engineering intervent …