The principal component analysis (PCA) is a well-known undirected method for reducing the number of variables used in further analyses. This is also called dimensionality-reduction. In some projects, you could have hundreds, maybe even thousands of input variables. Using all of them for can input to clustering algorithm could lead to enormous time needed to train the model. However, many of those input variables that might vary together, might have some association.

PCA starts again with the hyperspace, where each input variable defines one axis. PCA searches for a set of new axes, a set of new variables, which should be linearly uncorrelated, called the principal components. The principal components are calculated in such a way that the first one includes the largest possible variability of the whole input variable set, the second the second largest, and so on. The calculation of the principal components is derived from linear algebra. The principal...