Uses machine learning and statistical analysis to determine whether vehicle related factors contribute to the severity of road accidents. Conducted using Victorian road crash datasets as part of the COMP20008 subject assignment at the University of Melbourne

📊 Project overview 📊

This project entailed us analysing various Victorian Crash Datasets to determine correlations with specific factors and accident severity

🔧 Contributions 🔧

My main contribution to this project was to find any association between vehicle factors seen in vehicles.csv, and respective accident severities. This involved:

• Preprocessing and cleaning vehicle dataa
• Forming meaningful vehicle features
• Testing associtations using machine learning models

🧠Techniques used🧠

• Feature importance analysis
• Decision Tree Regressors
• Cross validation
• Coefficient of determination R^2
• Seaborn for visual comparisons

📌Key findings📌

• No significant association between the level of damage to a vehicle and the severity of an accident, however motorbikes had the least comparative association. This is understandable to the real world, as bikers are exposed in a crash, so severity of an accidnet is dependent on how bad the biker is hit, rather than a car which encases the victim
• Lexus being the safest car brand in terms of accident severity, while Ford and Holden are the least safe