Run Anomaly Detection
Compare Multiple Methods
Combine multiple methods using voting. Select 2+ methods below, set minimum votes, then create ensemble.
Evaluate Method Performance
Computes Precision, Recall, F1 using sanctioned athletes as ground truth
Combine multiple methods using voting. Select 2+ methods below, set minimum votes, then create ensemble.
Athlete Career Profile
Search for an athlete to view their career progression and performance distribution.
No athletes found.
Outliers flagged using Z-Score (threshold: 1.5σ). Other methods available in the Compare/Evaluate tabs.
Career Progression
Performance Distribution (Box Plot)
Performance Frequency Distribution
Scrape World Athletics Data
Scrape performance data from World Athletics database. Note: This may take several minutes.
Scraping Status
Scraping Log
Cluster Performances
Group performances into clusters to reveal natural performance tiers and temporal patterns.
Regimes
—
Performances
—
Athletes
—
Flagged
—
Method
—
Inertia
—
Career Span vs Max Regime Jump
Short career + large tier jump (top-left red zone) signals suspicious development. Orange = flagged; blue = clean.
Best Time vs Year Peak Reached
Career-best time vs. year peak was first reached. Lower-left (fast + early) = suspicious. Orange = flagged; clean athletes hidden by default — toggle via legend.
Performance Regime Summary
| Regime | Size | Mean Time | Best Time | Worst Time | Std Dev | Year Range |
|---|