Robert837-sys/kenya-road-safety-project

An AI prediction system aiming to predict and control Kenya road accidents. # Kenya Road Accident Analytics & Risk Prediction System > **A complete end-to-end data science pipeline analyzing NTSA Kenya road accident data to uncover patterns, predict fatality risk, and provide actionable insights for road safety authorities.** --- ## Table of Contents - Project Overview - Problem Statement - Research Questions - Project Structure - Dataset Description - Installation & Setup - How to Run the Project - Notebooks Overview - SQL Files Overview - Machine Learning Models - Key Findings - Power BI Dashboard - Technologies Used - Project Results - Recommendations - Limitations - References --- ## Project Overview Kenya's roads are among the most dangerous in sub-Saharan Africa. The National Transport and Safety Authority (NTSA) reports that road accidents claim over **3,000 lives every year** and injure tens of thousands more. The economic burden exceeds **KSh 300 billion annually**. This project builds a complete data-driven road accident analytics and risk prediction system using real NTSA accident records. It covers the full data science lifecycle — from raw data ingestion through machine learning to an interactive Power BI dashboard — specifically designed to support evidence-based road safety decision making by NTSA and Kenya's county governments. --- ## Problem Statement > _"Kenya experiences thousands of road accidents annually, yet traffic management > authorities lack a centralized data-driven tool to identify high-risk routes, > peak accident times, and key contributing factors. This project analyzes Kenyan > road accident data to uncover patterns, predict accident-prone conditions, and > provide NTSA and county governments with an actionable dashboard to guide road > safety interventions."_ --- ## Research Questions | # | Research Question | Answered By | | ---- | --------------------------------------------------- | ----------------- | | RQ1 | Which counties and roads have the mos …