Two Complementary Classification Methods for Designing a Concept Lattice from Interval Data

This paper holds on the application of two classification methods based on formal concept analysis (FCA) to interval data. The first method uses a similarity between objects while the second considers so-called pattern structures. We deeply detail these methods in order to show their close links. This parallel study helps understanding complex data with concept lattices. We explain how the second method obtains same results and how to handle missing values. Most importantly, this is achieved in full compliance with the FCA-framework, and thus benefits from existing and efficient tools such as algorithms. Finally, an experiment on real-world data in agronomy has been carried out for decision helping in agricultural practices.

• Type: Journal
• Year: 2010
• Where: FOIKS
• Authors: Mehdi Kaytoue-Uberall, Zainab Assaghir, Nizar Messai, Amedeo Napoli
• Keywords: Classification Methods, FOIKS 2010, Formal Concept, Information Technology, So-called Pattern Structures