Mike Sips

Selecting good views of high-dimensional data using class consistency

Many visualization techniques involve mapping high‐dimensional data spaces to lower‐dimensional views. Unfortunately, mapping a high‐dimensional data space into a scatterplot involves a loss of information; or, even worse, it can give a misleading picture of valuable structure in higher dimensions. In this paper, we propose class consistency as a measure of the quality of the mapping. Class consistency enforces the constraint that classes of n–D data are shown clearly in 2–D scatterplots. We propose two quantitative measures of class consistency, one based on the distance to the classs center of gravity, and another based on the entropies of the spatial distributions of classes. We performed an experiment where users choose good views, and show that class consistency has good precision and recall. We also evaluate both consistency measures over a range of data sets and show that these measures are efficient and robust.

RecMap: Rectangular Map Approximations

In many application domains, data is collected and referenced by its geospatial location. Nowadays, different kinds of maps are used to emphasize the spatial distribution of one or more geospatial attributes. The nature of geospatial statistical data is the highly nonuniform distribution in the real world data sets. This has several impacts on the resulting map visualizations. Classical area maps tend to highlight patterns in large areas, which may, however, be of low importance. Cartographers and geographers used cartograms or value-by-area maps to address this problem long before computers were available. Although many automatic techniques have been developed, most of the value-by-area cartograms are generated manually via human interaction. In this paper, we propose a novel visualization technique for geospatial data sets called RecMap. Our technique approximates a rectangular partition of the (rectangular) display area into a number of map regions preserving important geospatial constraints. It is a fully automatic technique with explicit user control over all exploration constraints within the exploration process. Experiments show that our technique produces visualizations of geospatial data sets, which enhance the discovery of global and local correlations, and demonstrate its performance in a variety of applications

Visual data mining in large geospatial point sets

Visual data-mining techniques have proven valuable in exploratory data analysis, and they have strong potential in the exploration of large databases. Detecting interesting local patterns in large data sets is a key research challenge. Particularly challenging today is finding and deploying efficient and scalable visualization strategies for exploring large geospatial data sets. One way is to share ideas from the statistics and machine-learning disciplines with ideas and methods from the information and geo-visualization disciplines. PixelMaps in the Waldo system demonstrates how data mining can be successfully integrated with interactive visualization. The increasing scale and complexity of data analysis problems require tighter integration of interactive geospatial data visualization with statistical data-mining algorithms.

CircleView: a new approach for visualizing time-related multidimensional data sets

This paper introduces a new approach for visualizing multidimensional time-referenced data sets, called Circle View. The Circle View technique is a combination of hierarchical visualization techniques, such as treemaps [6], and circular layout techniques such as Pie Charts and Circle Segments [2]. The main goal is to compare continuous data changing their characteristics over time in order to identify patterns, exceptions and similarities in the data.To achieve this goal Circle View is a intuitive and easy to understand visualization interface to enable the user very fast to acquire the information needed. This is an important feature for fast changing visualization caused by time related data streams. Circle View supports the visualization of the changing characteristics over time, to allow the user the observation of changes in the data. Additionally it provides user interaction and drill down mechanism depending on user demands for a effective exploratory data analysis. There is also the capability of exploring correlations and exceptions in the data by using similarity and ordering algorithms.

Pixel based visual data mining of geo-spatial data

Abstract In many application domains, data is collected and referenced by its geo-spatial location. Spatial data mining, or the discovery of interesting patterns in such databases, is an important capability in the development of database systems. A noteworthy trend is the increasing size of data sets in common use, such as records of business transactions, environmental data and census demographics. These data sets often contain millions of records, or even far more. This situation creates new challenges in coping with scale. For data mining of large data sets to be effective, it is also important to include humans in the data exploration process and combine their flexibility, creativity, and general knowledge with the enormous storage capacity and computational power of todays computers. Visual data mining applies human visual perception to the exploration of large data sets. Presenting data in an interactive, graphical form often fosters new insights, encouraging the formation and validation of new hypotheses to the end of better problem-solving and gaining deeper domain knowledge. In this paper we give a short overview of visual data mining techniques, especially for analyzing geo-spatial data. We provide examples for effective visualizations of geo-spatial data in important application areas such as consumer analysis and census demographics.

Pixnostics: Towards Measuring the Value of Visualization

During the last two decades a wide variety of advanced methods for the visual exploration of large data sets have been proposed. For most of these techniques user interaction has become a crucial element, since there are many situations in which a user or an analyst has to select the right parameter settings from among many or select a subset of the available attribute space for the visualization process, in order to construct valuable visualizations that provide insight, into the data and reveal interesting patterns. The right choice of input parameters is often essential, since suboptimal parameter settings or the investigation of irrelevant data dimensions make the exploration process more time consuming and may result in wrong conclusions. In this paper we propose a novel method for automatically determining meaningful parameter- and attribute settings based on the information content of the resulting visualizations. Our technique called Pixnostics, in analogy to Scagnostics (Wilkinson et al., 2005), automatically analyses pixel images resulting from diverse parameter mappings and ranks them according to the potential value for the user. This allows a more effective and more efficient visual data analysis process, since the attribute/parameter space is reduced to meaningful selections and thus the analyst obtains faster insight into the data. Real world applications are provided to show the benefit of the proposed approach

Information Visualization: Scope, Techniques and Opportunities for Geovisualization

Never before in history has data been generated at such high volumes as it is today. Exploring and analyzing the vast volumes of data has become increasingly difficult. Information visualization and visual data mining can help to deal with the flood of information. The advantage of visual data exploration is that the user is directly involved in the data mining process. There are a large number of information visualization techniques that have been developed over the last two decades to support the exploration of large data sets. In this article, we provide an overview of information visualization and visual data mining techniques, and illustrate them using a few examples. We show that an application of information visualization methods provides new ways of analyzing geography ‐ related data.

43 Visual Data-Mining Techniques*

Never before in history has data been generated at such high volumes as it is today. Exploring and analyzing the vast volumes of data has become increasingly dicult. Information visualization and visual data mining can help to deal with the flood of information. The advan- tage of visual data exploration is that the user is directly involved in the data mining process. There are a large number of information visualiza- tion techniques that have been developed over the last two decades to support the exploration of large data sets. In this paper, we propose a classification of information visualization and visual data mining tech- niques based on the data type to be visualized, the visualization technique, and the interaction technique. We illustrate the classification using a few examples, and indicate some directions for future work.

PixelMaps: a new visual data mining approach for analyzing large spatial data sets

PixelMaps are a new pixel-oriented visual data mining technique for large spatial datasets. They combine kernel-density-based clustering with pixel-oriented displays to emphasize clusters while avoiding overlap in locally dense point sets on maps. Because a full evaluation of density functions is prohibitively expensive, we also propose an efficient approximation, Fast-PixelMap, based on a synthesis of the quadtree and gridfile data structures.

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