Spurious Correlations
Source: tylervigen.com
This site serves as an interactive educational demonstration of 'data dredging,' showing how large datasets can produce statistically significant correlations that are completely coincidental. It challenges the assumption that correlation implies causation by displaying hundreds of randomly matched variables that share trend-lines over time.
Categorisation
Stage 3: Architect (11-13+)Pattern Recognition & Data AnalysisScreen-Assisted / HybridMixed/Multi-subject
Props & Materials Required
- •Large datasets
- •Statistical software (e.g. R, Python/Pandas)
- •Concept of correlation vs causation
Why It Works & Notes
Parent Note:
Statistics, correlation vs. causation, p-value limitations, data dredging, and the dangers of overfitting in data analysis.