Visual data analysis involves both open-ended and focused exploration. Manual chart specification tools support question answering, but are often tedious for early-stage exploration where systematic data coverage is needed. Visualization recommenders can encourage broad coverage, but irrelevant suggestions may distract users once they commit to specific questions. We present Voyager 2, a mixed-initiative system that blends manual and automated chart specification to help analysts engage in both open-ended exploration and targeted question answering. We contribute two partial specification interfaces: wildcards let users specify multiple charts in parallel, while related views suggest visualizations relevant to the currently specified chart. We present our interface design and applications of the CompassQL visualization query language to enable these interfaces. In a controlled study we find that Voyager 2 leads to increased data field coverage compared to a traditional specification tool, while still allowing analysts to flexibly drill-down and answer specific questions.

Effective small multiple displays are created by partitioning a visualization on variables that reveal interesting conditional structure in the data. We propose a method that automatically ranks partitioning variables, allowing analysts to focus on the most promising small multiple displays. Our approach is based on a randomized, non-parametric permutation test, which allows us to handle a wide range of quality measures for visual patterns defined on many different visualization types, while discounting spurious patterns. We demonstrate the effectiveness of our approach on scatterplots of real-world, multidimensional datasets.

General visualization tools typically require manual specification of views: analysts must select data variables and then choose which transformations and visual encodings to apply. These decisions often involve both domain and visualization design expertise, and may impose a tedious specification process that impedes exploration. In this paper, we seek to complement manual chart construction with interactive navigation of a gallery of automatically-generated visualizations. We contribute Voyager, a mixed-initiative system that supports faceted browsing of recommended charts chosen according to statistical and perceptual measures. We describe Voyager's architecture, motivating design principles, and methods for generating and interacting with visualization recommendations. In a study comparing Voyager to a manual visualization specification tool, we find that Voyager facilitates exploration of previously unseen data and leads to increased data variable coverage. We then distill design implications for visualization tools, in particular the need to balance rapid exploration and targeted question-answering.

Bar charts are one of the most common visualization types. In a classic graphical perception paper, Cleveland & McGill studied how different bar chart designs impact the accuracy with which viewers can complete simple perceptual tasks. They found that people perform substantially worse on stacked bar charts than on aligned bar charts, and that comparisons between adjacent bars are more accurate than between widely separated bars. However, the study did not explore why these differences occur. In this paper, we describe a series of follow-up experiments to further explore and explain their results. While our results generally confirm Cleveland & McGill's ranking of various bar chart configurations, we provide additional insight into the bar chart reading task and the sources of participants' errors. We use our results to propose new hypotheses on the perception of bar charts.

We introduce a method (Scagnostic time series) and an application (TimeSeer) for organizing multivariate time series and for guiding interactive exploration through high-dimensional data. The method is based on nine characterizations of the 2D distributions of orthogonal pairwise projections on a set of points in multidimensional euclidean space. These characterizations include measures, such as, density, skewness, shape, outliers, and texture. Working directly with these Scagnostic measures, we can locate anomalous or interesting subseries for further analysis. Our application is designed to handle the types of doubly multivariate data series that are often found in security, financial, social, and other sectors.

In Wilkinson et al. [2011] we introduced a new set-covering random projection classifier that achieved average error lower than that of other classifiers in the Weka platform. This classifier was based on an L ∞ norm distance function and exploited an iterative sequence of three stages (projecting, binning, and covering) to deal with the curse of dimensionality, computational complexity, and nonlinear separability. We now present substantial changes that improve robustness and reduce training and testing time by almost an order of magnitude without jeopardizing CHIRP's outstanding error performance.

We introduce a classifier based on the L-infinity norm. This classifier, called CHIRP, is an iterative sequence of three stages (projecting, binning, and covering) that are designed to deal with the curse of dimensionality, computational complexity, and nonlinear separability. CHIRP is not a hybrid or modification of existing classifiers; it employs a new covering algorithm. The accuracy of CHIRP on widely-used benchmark datasets exceeds the accuracy of competitors. Its computational complexity is sub-linear in number of instances and number of variables and subquadratic in number of classes.

An ongoing challenge for information visualization is how to deal with over-plotting forced by ties or the relatively limited visual field of display devices. A popular solution is to represent local data density with area (bubble plots, treemaps), color (heatmaps), or aggregation (histograms, kernel densities, pixel displays). All of these methods have at least one of three deficiencies:1) magnitude judgments are biased because area and color have convex downward perceptual functions, 2) area, hue, and brightness have relatively restricted ranges of perceptual intensity compared to length representations, and/or 3) it is difficult to brush or link to individual cases when viewing aggregations. In this paper, we introduce a new technique for visualizing and interacting with datasets that preserves density information by stacking overlapping cases. The overlapping data can be points or lines or other geometric elements, depending on the type of plot. We show real-dataset applications of this stacking paradigm and compare them to other techniques that deal with over-plotting in high-dimensional displays.

We introduce a mathematical framework, based on the L∞norm distance metric, to describe human interactions in a visual data mining environment. We use the framework to build a classifier that involves an algebra on hyper-rectangles. Our classifier, called VisClassifier, generates set-wise rules from simple gestures in an exploratory visual GUI. Logging these rules allows us to apply our analysis to a new sample or batch of data so that we can assess the predictive power of our visual-processing motivated classifier. The accuracy of this classifier on widely-used benchmark datasets rivals the accuracy of competitive classifiers.

We introduce a method for organizing multivariate displays and for guiding interactive exploration through high-dimensional data. The method is based on nine characterizations of the 2D distributions of orthogonal pairwise projections on a set of points in multidimensional Euclidean space. These characterizations include such measures as density, skewness, shape, outliers, and texture. Statistical analysis of these measures leads to ways for 1) organizing 2D scatterplots of points for coherent viewing, 2) locating unusual (outlying) marginal 2D distributions of points for anomaly detection, and 3) sorting multivariate displays based on high-dimensional data, such as trees, parallel coordinates, and glyphs.

We developed a testbed containing: real time data from over 830 highway traffic sensors in the Chicago region, data about weather, and text data about events that might affect traffic. The goal was to detect in real time interesting changes in traffic conditions. Given the size and complexity of the data, we choose to build a large number of separate baseline models. We built a separate baseline for each hour in the day, for each day in the week, and for every 2 or 3 traffic sensors, resulting in over 42,000 separate baseline models. We also built a baseline engine to build the necessary baselines automatically. We modified an open source scoring engine to process in real time each new sensor reading, update the appropriate feature vectors, score the updated feature vectors using the baseline models, and send out real time alerts when deviations from the baselines were detected.