The inexpensive storage and the ubiquity of digital systems, has resulted in an increasing number of entities including cities, federal governments, retailers, scientific organizations, NGOs, and even individuals are amassing huge databases, approaching terabytes and petabytes. Therefore, the need for data mining, i.e. extracting knowledge from the data in the form of useful and interesting models and trends, has become more important than ever.

Kernel methods enable application of classical linear data analysis techniques such as used in regression, classification, clustering, and principal component analysis, to non-linear problems. The novelty of the kernel method for non-linear analysis is that the theory underlying the algorithm to which the kernel is applied does not change and can still be described in terms of linear analysis.