Principal component analysis (PCA)

An unsupervised linear transformation that produces new features, called principal components, in the directions of maximum variance in the data. It does not take into account which features matter for classification.

See: PCA versus a Random Forest