Decreasing occlusion and increasing explanation in interactive visual knowledge discovery

Abstract

Explanation and occlusion are the major problems for interactive visual knowledge discovery, machine learning and data mining in multidimensional data. This paper proposes a hybrid method that combines the visual and analytical means to deal with these problems. This method, denoted as FSP, uses visualization of n-D data in 2-D, in a set of Shifted Paired Coordinates (SPC). SPCs for n-D data consist of n/2 pairs of Cartesian coordinates, which are shifted relative to each other to avoid their overlap. Each n-D point is represented as a directed graph in SPC. It is shown that the FSP method simplifies the pattern discovery in n-D data, providing the explainable rules in a visual form with a significant decrease of the cognitive load for analysis of n-D data. The computational experiments on real data has shown its efficiency on both training and validation data.