Knowledge discovery is the process of identifying useful patterns from large data sets. There are two families of approaches to be used for knowledge discovery: clustering, when the classes of domain objects are not known; and inductive learning algorithms, when the classes are known and the goal is to construct a domain model useful to identify new unseen objects. Clustering algorithms have also been proposed to analyze the data when the classes are known. However, to our knowledge, inductive learning methods are not used to analyze the available data but only for prediction. What we propose here is a methodology, namely FTree, that uses a decision tree to analyze both the available data identifying patterns and some important aspects of the domain (at least from the domain’s part represented by the data at hand) such as similarity between classes, separability, characterization of classes and even some possible errors on data.

This research is partially funded by the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 689176 (SYSMICS project), the projects RASO (TIN2015-71799-C2-1-P) and RPREF (CSIC Intramural 201650E044) and the grants 2014-SGR-118 and 2014-SGR-788 from the Generalitat de Catalunya.

Peer reviewed