How NOT to Lie with Visualization by Bernice E

How NOT to Lie with Visualization by Bernice E. Rogowitz, Lloyd A. Treinish

Summary of Paper Discussion by Balamurugan Settu

The paper gives a good idea about how visualizations can sometimes mislead the users. What representations are appropriate for one’s visualization and the importance of how variations in the method of representing the data can significantly influence the user’s perception and interpretation of data? How the interpretation of data can be enhanced? Four color maps such as default, isomorphic, segmented and highlighting are discussed with respect to how the representation can influence the interpretation of data. Having the flexibility of increasing the number of dimension does not necessarily mean better representation. The appropriate use of color is also one major area where the visualizationdesigner needs to concentrate, E.g., If a blue and red object is placed behind green the blue object appears as bluish-green where as the red object appears yellow.

In order for accurate representation of data one has to understand the structure in the data. It is important to understand the relationship between data structure and visual representation. In nominal data, where the data are not ordered, objects should be different in representation. In ordinal data, where the data are ordered, perceptual discrimination of objects should be apparent in the representation. In interval data, equal steps in data value should appear in the representation. In ratio data, where values increase and decrease monotonically needs to be preserved of its characteristics in the representation. The kind of visualization used should also depend on the spatial frequency of the data. The balance of luminance and saturation variation in an isomorphic color map depends on the spatial frequency of the data. For low frequency data, additional levels in segmentation tend to increase the information provided in the visualization.

Solving some of these problems, the author comes out with his own approach namely, PRAVDA (Perceptual Rule-based Architecture for Visualizing Data Accurately). In this rule-based architecture, rules filter the choices offered to the user, based on principles of human perception, attention and color theory. The users are constrained to select only some color maps from a library of color maps that are based on perceptual rules inferred from the data type, data spatial frequency, visualization task and other design choices made by the user. The tool analyses the structure in the data to be visualized (ordinal, interval or ratio), gives the option for the user to select the goal of the visual representation (e.g., isomorphic, segmentation or highlighting) and constrains the user to select only those color maps that are appropriate for the task. Thus a rule-based system seems to work better than a tool-based system as the rules can always be extended whenever we learn something from areas of human perception and color theory. Future work can be extended to data without any know structure and complex data that may have combination of structures.