About the Project

Author

E. Pitzer

Github LinkedIn

The Scenario

As a Data Scientist in the healthcare sector, accessing real patient data for public portfolios is restricted by HIPAA regulations.

To demonstrate my forecasting capabilities without compromising patient privacy, I engineered a synthetic dataset that mimics the statistical properties of a real Oregon-based community health center.

Methodology

This project simulates a 6-year historical dataset (2020-2025) incorporating:

  1. Trend: A realistic 3% annual patient volume growth.
  2. Seasonality: Weighted factors for high-traffic months (August/December) and low-traffic months (February).
  3. Weekly Cycles: Daily variance accounting for clinic operating hours (Closed Sundays, half-day Saturdays).
  4. External Shocks: A programmed “structural break” in Q1 2020 to simulate the impact of COVID-19 lockdowns on elective care.

Tech Stack

This project uses a “Code-First” approach to analytics:

  • Python: Data generation and logic.
  • Statsmodels: SARIMA forecasting and Seasonal Decomposition.
  • Pandas: Time-series manipulation and resampling.
  • Quarto: Reproducible reporting and HTML publishing.
  • GitHub Actions: To automate monthly data updates.