Principal Component Analysis (PCA) in 3D Visual Environment
(Included in Miner3D Professional and Miner3D Enterprise) An advanced user may often need to process high dimensionality datasets. PCA is a great tool to get an overview of the data, to detect groupings, trends or outliers and to evaluate correlations among variables and to reduce data dimensionality without loss of information. Its methods allow you to prepare data subsets more suitable for data visualization and increase the chance of producing useful results from your analyses. PCA allows you to focus the key information providing a way of condensing the information contained in a number of original variables into a smaller number of principal components. High dimensionality information is often borne by a few important dimensions, or their combinations, while the rest contains measurement "noise" or other uninteresting data. PCA will identify these important dimensions and will select a "subspace" in the data that contains the most information, throwing away the remaining dimensions. PCA aims to finds a new set of axes (PCA Vectors) such that most of the variability of the data is contained in the first few dimensions. The PCs are independent and uncorrelated variables that explain the observed variability. Each PC is a linear combination of the original variables. PCA in Miner3D is implemented via Eigen decomposition of the covariance matrix.
