Towards More Clean Results in Data Visualization: A Weka Usability Experiment

Abstract

The development of Information Technologies has contributed to the increase and complexity of data, implying a greater diversity of mechanisms for knowledge extraction. This high data availability has made citizens increase their interest in analyzing data and making more informed decisions. Data mining is an intrinsically complex process that expert users generally use. The non-expert users are overwhelmed because they lack relevant techniques for analyzing and understanding these results. This proposal presents a usability experiment to evaluate the level of understanding of the results when applying classification techniques. The users worked with decision trees, one of the “friendliest” of existing patterns. We need to start focusing on new patterns for non-expert users from the results exposed.