Extreme events: blending PCA and ICA [33]

Go to: Summary | Previous | Next   
Bullet points include: Quantile-quantile box plot had peaks in four corners. One reason is that chart includes a mixture of distributions. Different pairs of sectors have different correlations hence different distributions. We can eliminate this effect by focusing on principal components. Orthogonal by construction. Hence all disjoint pairs of principal components have the same (i.e. zero) correlation

Contents | Prev | Next | Library

Desktop view | Switch to Mobile