Four Years of Running: A Personal Performance Analysis
I exported four years of GPS watch data and built a full analysis of my training patterns, race performance, and recovery trends — then turned it into an interactive Tableau dashboard.
Challenge
Raw GPS exports are messy: duplicate entries, missing elevation data, inconsistent distance units across devices. The first challenge was building a reliable cleaning pipeline before any analysis could happen.
Approach
Used Python (pandas) to clean and normalize ~1,400 activity files. Engineered features like training load, weekly mileage rolling averages, and pace-per-elevation-gain. Visualized in Tableau with filters by year, activity type, and race vs. training.
Outcome
Identified that my best race performances followed a specific taper pattern I hadn't consciously noticed. Adjusted my training plan accordingly — and ran a 4-minute PR at my next half marathon.
