DMP: Traffic Accident Severity Prediction in Austria

Context and methodology

This Data Management Plan (DMP) was created for the FAIR Data Science course (DaSt 2026) at TU Wien. The project investigates whether the severity of Austrian road traffic accidents (slight injury, serious injury, or fatality) can be predicted using a Random Forest Classifier trained on contextual features such as road type, weather, lighting, and time of day.

The input dataset is the Straßenverkehrsunfälle mit Personenschaden ab 2013 published by Statistik Austria (data.statistik.gv.at) under CC BY 4.0. Data is processed using Python 3 with pandas and scikit-learn, with a stratified 70/15/15% train/validation/test split.

Technical details

The DMP follows the FWF Data Management Plan template (01/2022), based on Science Europe Core Requirements. It covers data characteristics, metadata standards (Dublin Core), data quality control, storage and sharing strategy (TU Wien Research Data Repository, DOI), and legal/ethical aspects. All outputs are published under CC BY 4.0 (data/figures) and MIT licence (code). No personal or sensitive data is involved — the input dataset is fully anonymised by Statistik Austria.