From CS 294-10 Visualization Sp10

Jump to: navigation, search


The main story of my scatterplot visualization explains how the size of a country (in population) correlates with its pledge of aid per capita, and how these factors determine the total amount pledged by the country. The bar chart clarifies how countries and organizations rank in terms of absolute quantity pledged, and how different categories of donors (classified by geographic location and organization type) compare. Unfortunately, my visualization largely ignores what percentage of the total each donor contributes, since a World Total datapoint at (7 billion people, $2,422,202,996) would be an outlier that would exceed the space limitations of my chart.

On the scatterplot, I chose to encode the nominal datapoint identifiers using flags, which combine color and texture dimensions to create an integral dimension with moderate filtering interference and some redundancy. Although the interference is unfortunate, I think that the very compact and intuitive encoding (most people can instantly recognize and associate flags with countries) of the nominal values is worth the cost. In case people do not recognize a particular flag, there is a key that matches the flag icon to an English spelling of the country name.

The quantitative population and contribution per capita are encoded as position along common scale, which is a simple, high-accuracy comparison according to the Cleveland & McGill paper. The scatterplot realization allows the viewer to perceive the direction between pairs of points, in order to better understand the relationship between population and pledge-per-capita. A secondary benefit of using a scatterplot is that the total contribution of the country, which is the product of the population and per-capita contribution, can be be visualized as the area of a rectangle which has one corner at the origin, and the opposite corner at the datapoint. Unfortunately, area comparisons are not very accurate, therefore I decided to explicitly clarify these in a bar graph, which uses the more accurate length and position along a common axis dimensions to convey the quantitative total amount pledged. The quantities are sorted from highest to lowest, which correlates to distance from origin on the scatterplot, and thus should simplify matching datapoints on the two charts. The bar chart also classify donors into categories denoted by color, since color does not contain any inherent quantitative or ordinal connotations.

I think that my choice of using a 2-dimensional visualization is quite effective at revealing subtle correlations that may not be apparent in more standard (eg 1-dimensional) visualizations. For example, it is interesting to notice that Scandinavian countries are clustered along the Y-axis (low population, but high per-capita contribution), reflecting their priviledged economic position. In contrast, developing countries, including members of the the BRIC (Brazil, Russia, China) tend to stay near the Y-axis (higher population, lower contribution per capita). Finally, in terms of absolute donations, North American and Western European nations tend to dominate.

[add comment]
Personal tools