Non-Gaussian data requires robust PCA and factor analysis that ignores outliers and heavy tails. These methods include outlier-resistant covariance estimation and data distribution transformations.