/

Extreme events: blending PCA and ICA [3]

Go to: Summary | Previous | Next   
Bullet points include: Both are examples of ‘blind source separation’, aiming to identify ‘signals’ (i.e. sources / factors) that explain (observed) market behaviour. Principal Components Analysis (PCA). Seeks to identify the largest contributors to variance, i.e. magnitude of impact. ‘Signals’ maximise sum of variances of returns of each security within universe. Independent Components Analysis (ICA). Seeks to identify contributors to market behaviour that are meaningful. ‘Signals’ maximise independence, non-Normality and/or complexity

NAVIGATION LINKS
Contents | Prev | Next | Library


Desktop view | Switch to Mobile