The document describes a study comparing several dimensionality reduction and classification algorithms, including PCA, 2DPCA, (2D)2PCA, and 2DLDA, on a face recognition dataset. 2DPCA is an improvement on PCA that operates directly on image matrices rather than vectorizing them. (2D)2PCA is an improved version of 2DPCA that computes projections in two directions rather than one. Experiments show that (2D)2PCA consistently achieves the lowest test error rates of the methods compared across a variety of dimensions.