Applied Data Science Projects
Please view a collection of my individual and group projects with descriptions and code below. For access to reports and presentations, please emeail me at teeuwen.b1@gmail.com or message me on LinkedIn.
Analyzing E-Commerce Session Data to Predict Purchases
This project demonstrates how raw e-commerce session data can be transformed into clear, actionable insights about purchasing behavior. By combining exploratory analysis, feature engineering, and predictive modeling, the project identifies which user behaviors and traffic sources are most strongly associated with successful transactions and highlights where business interventions are likely to have the greatest impact.
North American renewable energy capacity
This project delivers a machine learning–based forecast of renewable energy generation in North America and demonstrates that, under realistic growth assumptions, existing and planned renewable capacity will not be sufficient to meet projected AI-driven data center electricity demand by 2028. The analysis quantifies the expected capacity shortfall and highlights where infrastructure planning and investment gaps exist.
Predicting Responder Outcomes In A Lung Cancer Drug Trial Using ML
This project demonstrates how clinically meaningful insights can be extracted from
imperfect trial data. By combining statistical testing and machine learning, the analysis
identified immune response and biomarker signals associated with treatment success,
evaluated relative drug efficacy across trial arms, and translated these findings into
practical guidance for FDA review and Phase II trial planning.
Forecasting Under Volatility: A Time-Series Approach to TSA Operations
This project demonstrates that TSA passenger demand is highly predictable despite external shocks and seasonal volatility. By identifying a robust forecasting approach and translating demand forecasts into workforce requirements, the analysis provides a practical framework for proactive staffing that reduces inefficiencies, stabilizes wait times, and improves operational readiness.
Forecasting by Product: An Interactive Time-Series Analysis of Wine Sales
This project demonstrates how interactive dashboards can be used to translate time-series forecasting results into business insight. By comparing multiple models across wine varietals, the analysis shows that forecasting strategies must be tailored to each product’s demand pattern and provides decision-makers with an intuitive way to explore uncertainty and model performance.
Agentic AI: An Applied Decision-Support Analysis for CRM Workflows
This project demonstrates how agentic AI systems can be evaluated as decision-support tools rather than purely predictive models. By separating the decision of when to act from how to act, the analysis shows how thresholded, two-stage designs enable controllable, precision-first deployment and make the tradeoffs between accuracy, coverage, and risk explicit for business decision-makers.
