![]() Particularly, subgroup relationships and ranges can be clearly observed, even as the number of data subgroups grows past two or three. Splatterplots allow a user to see trends and patterns in datasets even as the number of points grows large relative to the diagram size. Continuous zooming automatically reveals abstracted details since abstractions are determined in screen space. The details removed in a Splatterplot can be revealed through interaction mechanisms such as zooming and panning, enabling fast exploration of massive datasets. A screen space metric controls both the amount of detail in the smooth shapes and the display density of outlier glyphs, enforcing a limit on the amount of visual complexity. Dense regions are shown as smooth shapes, and outlying points are subsampled so they do not exceed a specified visual density. Splatterplots abstract point data so that the amount of information shown in any area of screen space is bounded. Our key idea is to limit the amount of visual complexity in any screen space area. In this paper we introduce Splatterplots, a novel presentation of point data that addresses weaknesses in scatter plots to better scale to larger datasets. Additionally, many analysis and visualization techniques, such as scatter plot matrices (SPLOMS), call for several different concurrent plots, further diminishing the amount of available screen space. Even if we decrease the size of the glyphs to be a single pixel, an increasing number of data sets contain many more points than there are pixels in a monitor. High density regions, large numbers of points, and multiple data sets all contribute to the overdraw problem. Overdrawing increases as the available drawing space per glyph decreases. This interferes with the viewer’s ability to group points perceptually and spot outliers. This overlap can become so severe that it is impossible to perceive the number of points in a given region of the scatter plot. Overdrawing occurs when the glyphs that are used to visualize data points overlap. Unfortunately, scatter plots become less effective as the overlap within points increases. These properties make scatter plots good for exploring data sets and communicating interesting findings. Additionally, scatter plots offer a means for comparing different data sets when plotted on the same axes. They can make outliers easy to identify because regions with higher density of points will be grouped perceptually. Scatter plots can display data trends and correlations between any two dimensions. SCATTER plots are a simple, intuitive and natural way of visualizing two dimensional point data. ![]() We show how splatterplots can be an effective alternative to traditional methods of displaying scatter data communicating data trends, outliers, and data set relationships much like traditional scatter plots, but scaling to data sets of higher density and up to millions of points on the screen. We present techniques that leverage the GPU for Splatterplot computation and rendering, enabling interaction with massive data sets. The resulting visualizations represent the dense regions of each subgroup of the dataset as smooth closed shapes and show representative outliers explicitly. We combine techniques for abstraction with with perceptually based color blending to reveal the relationship between data subgroups. Abstraction automatically groups dense data points into contours and samples remaining points. ![]() To address these issues, Splatterplots abstract away information such that the density of data shown in any unit of screen space is bounded, while allowing continuous zoom to reveal abstracted details. Overdraw obscures outliers, hides data distributions, and makes the relationship among subgroups of the data difficult to discern. Traditional scatter plots suffer from overdraw (overlapping glyphs) as the number of points per unit area increases. We introduce Splatterplots, a novel presentation of scattered data that enables visualizations that scale beyond standard scatter plots. ![]()
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |