Team · Team Project | Predictive Analytics
Accident Risk Prediction
A supervised machine-learning workflow that models historical road-incident data to estimate accident risk levels across environmental, temporal, and roadway conditions.
My contribution
I contributed to data preparation, feature work, model experiments, and framing outputs as risk indicators for stakeholders — not absolute predictions of individual events.
Python · Pandas · Scikit-learn · Feature engineering · Classification · Model evaluation · Imbalanced learning
Problem
Road safety interventions are often reactive due to a lack of granular risk scoring across combinations of weather, road geometry, and traffic timing.
Solution & architecture
Executed end-to-end tabular ML engineering — handling missing data, encoding categorical road/weather variables, engineering interaction features, and training classification/risk-scoring models.
Feature engineering & evaluation
Temporal bucketing (rush hour, night), weather severity encoding, and road-type interactions, with class-imbalance handling and precision-recall/ROC analysis across risk tiers.
My contribution
Team contribution across cleaning, feature engineering, experimentation, and stakeholder-facing interpretation of risk signals.
Impact
Identifies high-risk condition combinations to support preventative traffic advisories and resource allocation discussions.