Computational Techniques in Visual Analytics

Visual analytics approaches combine interactive visualisations with the use of computational techniques for data processing and analysis. Combining visualisation and computation has two sides. One side is computational support to visual analysis: outcomes of computations are intended to provide input to human cognition; for this purpose, they are represented visually. The other side is visual support to application of computational methods, which includes visual exploration of data properties for preparing data to computations, evaluation of computation outcomes, and comparison of results of different runs of computational techniques. This chapter focuses in more detail on the computational support to visual analysis. A major common purpose for using computational methods in visualisation is enabling an overview of voluminous and complex data. The general approaches are spatialisation, which is achieved by means of data embedding techniques, and grouping, which is achieved using clustering algorithms. Distance functions are auxiliary computational techniques used both for data embedding and clustering. They provide numeric assessments of the dissimilarity between data items. Another class of auxiliary techniques is feature selection. The chapter describes the use of data embedding and clustering. By example of clustering, the general principles of visual analytics support to application of computational methods are demonstrated. Topic modelling is a special group of data embedding techniques originally designed for textual data but applicable also to other kinds of data. The main ideas, properties, and uses of the topic modelling methods are discussed in a separate section.