S-m-i-l-e-e-e/kaduna-traffic-DSS

Hybrid ML + rule-based Decision Support System for predicting road traffic crash severity across Kaduna State LGAs. # Kaduna State Road Traffic Crash DSS A **Hybrid Machine Learning Decision Support System (DSS)** for predicting road traffic crash severity, fatalities, and injuries across Local Government Areas (LGAs) in **Kaduna State, Nigeria**. Built as a Final Year Project by **Mahmud Fareedah Lawal** under the supervision of **Dr. Ayodeji S. Makinde**. ## What This System Does - **Predicts** accident outcomes (accident count, killed, injured) for **8 core LGAs** using a hybrid ML + rule-based engine - **Classifies** crash severity as **Mild**, **Moderate**, or **Severe** - **Visualizes** historical trends, correlations, and sector-level analytics - **Maps** accident intensity using interactive choropleth maps (all **23 LGAs**) - **Validates** inputs against realistic thresholds derived from training data ## Hybrid Architecture **1. Input Validation**: Hard limits + warnings on all inputs **2. Rule-Based Fallback**: LGA-specific baselines + quarter modifiers + domain-knowledge weights **3. Machine Learning**: Random Forest for severity classification (95.6% accuracy) and accident regression (MAE 2.8) **4. Decision Fusion**: Auto-fallback to rule-based when ML confidence < 61% or uncertainty is high