PCA is a multivariate data analysis technique used to take a large number of data streams and express the patterns and correlations found in the data space in a smaller number of data streams called principal components. Each principal component data stream is a linear combination of the original data streams and each extracts the maximum possible variance from the original data set that has not already been expressed in a prior principal component. The principal components are created such that the 1^{st} principal component will account for the most variance in the original data streams, the 2^{nd} principal component will account for less variance and so forth. The PCA technique actually produces as many principal components as there are data streams, however the first few components are the most important as they express the most variance found in the original data stream.