AT A GLANCE
Developed predictive models and an interactive dashboard to analyze accident severity and risk factors for human-driven and autonomous vehicles. Achieved ROC-AUC scores up to 0.83, identified top contributors like weather and lighting conditions, and reduced manual data analysis through a feature-rich Streamlit app. The solution supports ongoing safety research and informed decision-making.
Client information
CHALLENGE
The client needed deeper insights into accidents involving human-driven and autonomous vehicles to enhance safety research. They aimed to develop predictive models to assess accident frequency and severity, improving their vehicle safety solutions.
SOLUTION
Selected and preprocessed four key datasets (CRSS, FARS, and autonomous vehicle data) to ensure consistency and build predictive models assessing accident severity and frequency for human-driven and autonomous vehicles. Developed an interactive Streamlit app for exploring accident data, including real-time risk dashboards and exploratory data analysis (EDA). Conducted correlation analysis to guide feature selection and delivered comprehensive documentation for model retraining and app usage.
IMPACT
TOOLS
Ready to achieve measurable results? Tell us about your challenges, and we’ll show you how we can help
