Correlation is Not Causation

Tags:

An excellent article on how to detect and avoid the most common mistake when interpreting time-series.

Tom Fawcett (@tomeff) from Silicon Valley Data Science, a data science consultancy, demonstrates how to transform two absolutely random time series in order to have them appear highly correlated.

Several common techniques to detect and remove trends are presented. A must-read if you were ever faced with the “correlation is not causation” mantra.