Mapping without visualizing local default is nonsense
Sylvain Lespinats, Michael J Marie Aupetit · The European Symposium on Artificial Neural Networks · 2010
High-dimensional data sets are often embedded in two-dimensional spaces so as to visualize neighborhood relationships. When the map is effective (i.e. when short distances are preserved) it is a powerful way to help an analyst to understand the data set. But, mappings most often show defaults and the user is then led astray. According to this notion, a mapping should not be considered when its overall quality is not good enough. Many imperfect mappings can however be exploited by informing the user of the nature and level of defaults. In this work, we propose to visualize local indices trustworthiness and continuity for that purpose.