📌 Project Type: Academic Research – Environmental & Epidemiological Modeling

📍 Published in: Science of the Total Environment (STOTEN)

👨‍🔬 Role: Co-Author – Stepwise Regression & Data Support

📄 Read Full Paper →

This research study examined how demographics and air quality influenced COVID-19 infection and fatality rates across counties in New York State during the pandemic's first wave. The study revealed that infection and death were highest near NYC, while fatality (deaths per infection) was paradoxically higher in rural areas.

✅ Stepwise Regression Modeling

Applied forward selection and backward elimination techniques to identify statistically significant predictors (demographic & environmental) for COVID-19 infection and fatality.

Helped determine which features most improved model accuracy (e.g., PM2.5, population age, distance to epicenter).

✅ Data Wrangling & Cleaning

Merged multiple datasets (census, pollution, and epidemiological data) at the county level.

Preprocessed variables for model readiness: normalization, missing value handling, and encoding.

✅ Feature Impact Interpretation

Analyzed how variable inclusion/exclusion altered regression accuracy and output.

Supported result validation to ensure models aligned with observed cluster behaviors.

PM2.5 and distance to NYC were major predictors of infection spread

Fatality was more associated with elderly population and long-term pollution exposure

Spatial and demographic segmentation is crucial for targeted public health response

Model interpretability helped explain why certain rural areas had high fatality despite low infection

This project shows my ability to:

Select and justify data modeling techniques

Build and interpret multivariate regression models

Understand feature importance & business impact

Handle real-world public datasets at scale

Applicable for:

Health analytics, churn modeling, marketing attribution, or KPI drivers

Python • pandas • stepwise regression (forward/backward) • multivariate analysis • data cleaning • clustering analysis • scientific writing