Home

PlaneSEEKER

       The classification process executed by PlaneSEEKER is based on the application of constructive induction in machine learning. The main feature of the program relies on the use of standard binary classificatory (capable, by definition, for classification of objects belonging to only two well separated categories) for the multicategory classification. This approach is based on the exhaustive dichotomization of a dataset, i.e. creation of n*(n-1)/2 subsets, if the original (source) dataset contains n categories. The constructive induction mechanism can – in many cases – supply interesting and valuable information about the inherent content of the analyzed dataset, for example it can points out cases (objects) creating outlying examples. This feature seems to be indispensable for medical diagnoses, when very similar, or even the same symptoms describe the new (unknown) form of a disease, having already fixed properties.