Classification based on rules and thyroids dysfunctions (Q2711701)

Common characteristics of ill-structured domains are: (1) heterogeneous data matrices - the variables used to describe the objects may be either quantitative or qualitative; (2) additional knowledge of the domain structure is available - usually it is qualitative knowledge relative to the overall structure of the domain (relationships among variables, classification goals, etc.); (3) partial and non-homogeneous knowledge - experts use to deal with great amounts of implicit knowledge. Although additional expert knowledge on the domain structure is always available, construction of complete knowledge bases to be used in diagnostic-oriented ~systems is almost unreachable for ill-structured domains due to the complexity of these kinds of domains. In such a way, classification in ill-structured domains is a difficult problem for the actual statistical and artificial intelligence techniques, because of the intrinsic characteristics of those domains.NEWLINENEWLINENEWLINEClassification based on rules is the present paper's proposal to overcome the limitations of statistics and artificial intelligence techniques referred to in this particular context. An application of the classification based on rules to a set of real data is presented. The data base is on thyroid function and was provided by a hospital from Zagreb (Croatia) covering a period of two years.