In Principal Component Analysis, the dimensionality defines how many ortogonal movements define a given threshold of variance.