Statistical methods for variables space reduction in cephalometric studies

The medical databases usually contain records for a large number of parameters, being difficult to interpret properly; Data Mining is an analytical technique used to explore large quantities of data, in order to identify consistent patterns and systematic relationships between variables, and to validate these results by applying them to new data sets. The Principal Components Analysis is a specific technique in this field, used to extract the smallest number of components from a large collection of variables, preserving as much as possible the information contained in the original data set. We present and discuss in this paper the theoretical background of this technique and a practical application in the specific case of cephalometric studies, which involves a large number of measurements on teleradiographies, being a perfect choice for this type of analysis.