Graphical representation and exploratory visualization for Decision Trees in the KDD process

This article presents a proposal of representation and scheme of exploratory visualization for Decision Trees (D-T) in the KDD (Knowledge Discovery in Database) process, specifically in the data mining stage. With this, the improvement of the understandability of the internal operation of the model is pursued. This exploratory visualization is based on the wellknown technique named Treemap (maps of trees), that allows representing hierarchical structures like the D-T, being used grids to represent the nodes of the D-T. The proposed visualization represents the number of instances or weight associated to a node with a scale of colors in degradation. In this way it is managed to heighten the rules of the D-T in a 2D and 3D graphical representation of this visualization. Additionally, a metadata structure is defined to be able to represent the graphical visualization of the D-T, and with this to allow users the interoperability of the model with other visualization techniques and later on, to add interaction mechanisms to it. Finally, a first attempt of subjective evaluation, based on for criteria, of the proposed visualization, is made. In this sense, this work is a first step to introduce new schemes of visualizations that allow specifically understanding how the data mining models work internally.