Author(s)

A F Gualtierotti

Abstract

Data mining a large health insurance data- base: a follow-up A.F. Gualtierotti IDHEAP, Switzerland Abstract Reference Gualtierotti [1] describes a data mining problem and a method to tackle it, but says nothing about results as, at the time of publication, none were available. This follow-up contains a description of the first results that were gathered using that method, but only as they inform on data mining particulars rather than on the questions of substance pertaining to the associated data mining project. 1 Introduction The database in question has a number of characteristics that are important for understanding the nature and possible interest of the data mining problem considered. These features are: The database is an administrative database, and as such contains fairly complex data that almost always require interpretation and then validation. For such a base, for instance, if sex can be processed similarly to a bar of soap, a psychiatric patient, for example, is harder to identify and requires a necessarily somewhat arbitrary definition and possibly the intervention of a psychiatrist (especially when the analyses must be accepted by doctors). Thus the objects of the analysis are fuzzier than those of traditional data mining.

Keywords