A1-RobertCarroll
From CS294-10 Visualization Fa07
Contents |
[edit] Good Visualization
From Lang, Kenneth R. Sun Earth and Sky, Second Edition, p. 107.
[edit] Explanation
This visualization was found in a textbook used in an undergraduate course on the sun and solar system. The image is a composite of multiple images of the sun take over the course of a solar eclipse. The purpose of the visualization is to concisely show the progression of an eclipse, as well as simply being a visual tool to capture the readers attention. This image is a multi-exposure photograph of a total solar eclipse. The photograph was presumably made by exposing the same negative multiple times as the eclipse progressed, although a similar effect could be achieved be compositing multiple independent photographs.
[edit] Deconstruction
The data set for this visualization is a bit less clear than in more traditional types of visualizations. On a one level the data set can be thought of as a set of individual camera exposures, that is of light rays hitting a negative. The individual exposures are encoded additively into a single image. On a higher level the data set is the motion of the sun across the sky, and the moon relative to the sun. The 3d positions of the objects are encoded in by the positions in the image through a perspective projection.
[edit] Critique
This image compresses the timeline of the event into a single visualization, allowing the viewer to more easily see the progression without watching a video or looking at multiple photographs. The image accentuates the dynamic factors present in a sequence of single-exposure images, that is the position of the sun/moon pair in the sky and the relative position of the moon over the sun, and minimizes the redundancy in the rest of the frame. At first glance it may seem trivial to make an image like this, but much care had to be taken to ensure each exposure let in the proper amount of light, and that the eclipse moved the proper amount between each exposure while knowing ahead of time that one exposure would capture the apex.
[edit] Bad Visualization
From The New York Times, Sept. 4 2007, p. A23.
[edit] Explanation
This chart taken from the New York Times attempts to show various metrics of the conditions in Iraq over the last few years. The purpose of the table is to illustrate the current state of Iraq in terms of military, violence, and civil infrastructure. The chart also shows which conditions are favorable and unfavorable, from the U.S. perspective.
[edit] Deconstruction
The data set for this visualization are the values of various metrics of progress in Iraq take yearly for the last five years. Rows in the table are used to encode which metric is being displayed, while the columns are used to encode the year. Values are encoded with a numerical value as well as a grayscale value with lighter shades being more favorable conditions. The data set consists of 18 metrics taken of 5 years, for 90 total entries. The table is displayed superimposed on an image of a clipboard and lined paper, which provide little information, but are presumable included to make the data seem more accessible.
[edit] Critique
Despite the effort that went in to neatly laying out the data, and the sheer amount of space used to display it, overall trends in the data do not pop out to the reader immediately. The grayscale encoding is the only aspect of the visualization that goes beyond a standard numerical table, and although it does help to some degree, there is lots of room for improvement. The image seems to give the impression that the different metrics are more closely related than they really are. Also, unlike any of the other rows, the first row contains two (albeit related) metrics instead of one. This seems like a particularly bad decision as it breaks the flow of the chart.
[edit] Redesign
In my redesign I chose to take display each metric as a scatterplot with the x-axis encoding time and the values encoded on the y-axis. This separation makes it more clear that much of the information being displayed isn't closely related. Also, positional encoding of the values is cleared than using grayscale, allowing trends top pop out more clearly. The plots are separated into two groups depending on whether higher values correspond to favorable or unfavorable conditions. The visualization remains in black in white since it is meant for a newspaper.
