Principal Component Analysis (PCA) is a statistical/unsupervised machine learning technique that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of linearly uncorrelated variables called principal components, thereby finding the maximum directions of variances in the dataset (along the principal components).
This can be used for (linear) dimensionality reduction (only a few dominant principal components captures almost all of the variance in a dataset most of the time) and visualization (in 2D) of datasets having many dimensions. One application of PCA is eigenfaces, to find a set of faces that can (theoretically) represent any face (as a linear combination ...
