Visual Interaction with Dimensionality Reduction: A Structured Literature Analysis.

Dimensionality Reduction (DR) is a core building block in visualizing multidimensional data. For DR techniques to be useful in exploratory data analysis, they need to be adapted to human needs and domain-specific problems, ideally, interactively, and on-the-fly. Many visual analytics systems have already demonstrated the benefits of tightly integrating DR with interactive visualizations. Nevertheless, a general, structured understanding of this integration is missing. To address this, we systematically studied the visual analytics and visualization literature to investigate how analysts interact with automatic DR techniques. The results reveal seven common interaction scenarios that are amenable to interactive control such as specifying algorithmic constraints, selecting relevant features, or choosing among several DR algorithms. We investigate specific implementations of visual analysis systems integrating DR, and analyze ways that other machine learning methods have been combined with DR. Summarizing the results in a "human in the loop" process model provides a general lens for the evaluation of visual interactive DR systems. We apply the proposed model to study and classify several systems previously described in the literature, and to derive future research opportunities.

Visual analysis of network traffic for resource planning, interactive monitoring, and interpretation of security threats.

The Internet has become a wild place: malicious code is spread on personal computers across the world, deploying botnets ready to attack the network infrastructure. The vast number of security incidents and other anomalies overwhelms attempts at manual analysis, especially when monitoring service provider backbone links. We present an approach to interactive visualization with a case study indicating that interactive visualization can be applied to gain more insight into these large data sets. We superimpose a hierarchy on IP address space, and study the suitability of Treemap variants for each hierarchy level. Because viewing the whole IP hierarchy at once is not practical for most tasks, we evaluate layout stability when eliding large parts of the hierarchy, while maintaining the visibility and ordering of the data of interest.

Visualization of geo-spatial point sets via global shape transformation and local pixel placement.

In many applications, data is collected and indexed by geo-spatial location. Discovering interesting patterns through visualization is an important way of gaining insight about such data. A previously proposed approach is to apply local placement functions such as PixelMaps that transform the input data set into a solution set that preserves certain constraints while making interesting patterns more obvious and avoid data loss from overplotting. In experience, this family of spatial transformations can reveal fine structures in large point sets, but it is sometimes difficult to relate those structures to basic geographic features such as cities and regional boundaries. Recent information visualization research has addressed other types of transformation functions that make spatially-transformed maps with recognizable shapes. These types of spatial-transformation are called global shape functions. In particular, cartogram-based map distortion has been studied. On the other hand, cartogram-based distortion does not handle point sets readily. In this study, we present a framework that allows the user to specify a global shape function and a local placement function. We combine cartogram-based layout (global shape) with PixelMaps (local placement), obtaining some of the benefits of each toward improved exploration of dense geo-spatial data sets.

Topological fisheye views for visualizing large graphs.

Graph drawing is a basic visualization tool that works well for graphs having up to hundreds of nodes and edges. At greater scale, data density and occlusion problems often negate its effectiveness. Conventional pan-and-zoom, multiscale, and geometric fisheye views are not fully satisfactory solutions to this problem. As an alternative, we propose a topological zooming method. It precomputes a hierarchy of coarsened graphs that are combined on-the-fly into renderings, with the level of detail dependent on distance from one or more foci. A related geometric distortion method yields constant information density displays from these renderings.

CartoDraw: a fast algorithm for generating contiguous cartograms.

Cartograms are a well-known technique for showing geography-related statistical information, such as population demographics and epidemiological data. The basic idea is to distort a map by resizing its regions according to a statistical parameter, but in a way that keeps the map recognizable. In this study, we formally define a family of cartogram drawing problems. We show that even simple variants are unsolvable in the general case. Because the feasible variants are NP-complete, heuristics are needed to solve the problem. Previously proposed solutions suffer from problems with the quality of the generated drawings. For a cartogram to be recognizable, it is important to preserve the global shape or outline of the input map, a requirement that has been overlooked in the past. To address this, our objective function for cartogram drawing includes both global and local shape preservation. To measure the degree of shape preservation, we propose a shape similarity function, which is based on a Fourier transformation of the polygons' curvatures. Also, our application is visualization of dynamic data, for which we need an algorithm that recalculates a cartogram in a few seconds. None of the previous algorithms provides adequate performance with an acceptable level of quality for this application. In this paper, we therefore propose an efficient iterative scanline algorithm to reposition edges while preserving local and global shapes. Scanlines may be generated automatically or entered interactively to guide the optimization process more closely. We apply our algorithm to several example data sets and provide a detailed comparison of the two variants of our algorithm and previous approaches.