A3-ChrisVolzKesavaMallela
From CS294-10 Visualization Fa07
Contents |
[edit] Area Code Map
[edit] Problem Statement
A casual look at the telephone area code map of United States reveals that there is no obvious allocation pattern of the codes. More importantly, the map hides area code information of large metropolitan areas of cities like New York, Los Angeles and Chicago at the expense of states like Wyoming and Idaho which have far lesser phone traffic than say NYC.
We want to redesign the map based on phone density of an area code. Since most households in United States have a phone line, we want to use population of the county as a weight of importance rather than geographical area.
[edit] Data Domain
We use area code data set from http://www.lincmad.com/areacode.html; historic population data is available from http://fisher.lib.virginia.edu/collections/stats/histcensus/
[edit] Technique
- Circles proportional to population
- Pixel area of area code proportion to population
[edit] Storyboard of Interface
We want to re-draw the map of area codes (mentioned in the above section) like the following California map. If we find that the scope is little too overwhelming for the purpose of this assignment, we might end up using proportional circles as overlays on the map.
[edit] Implementation
Current working version of implementation is here: http://people.ischool.berkeley.edu/~kesava/infoviz/AreaCodeCalif.html
Gathering and modifying the data took a great deal more time than we first imagined. Area codes do not adhere to any standard measured boundaries for population, so in order to gather population data for a specific area code we first had to determine what zip codes were contained within an area code and then sum population information for those zip codes to come up with a total for each Area Code. Once this task was complete we implemented a rough overlay map using Google's Map API where each area code was visualized as a circle, the radius of which was proportional to the size of the population contained within the area code. These overlays were then placed over the rough center of the area code's geographic area on the map.
The results were interesting though ultimately unsatisfying. Two key points did become clear. The first is that large metropolitan areas may be over represented due to the fact that larger populations will likely have more businesses and thus require a disproportionate number of telephone lines; at least in relation to less populated areas. It would be interesting to know how phone number density relates to population density. The second result that became immediately obvious was that the use of circular overlays is probably not the most effective way to visualize area code population densities due in part to the irregular boundaries of area codes and also because of the ability for a single area code to cover a very large, low population density area. Because of this, true centers of area codes for low population density areas may need to be recalculated.
Our original plan, and one that may still yet be pursued if we can find the time, was to create rectangular cartograms which were proportional to the population of the area code but arranged in rough approximation to the geographic area those area codes served. This probably would've resulted in a more interesting visualization where each area code would be given a proportional space and shape to its true nature. With our current visualization high density areas such as Los Angeles and the San Francisco Bay Area show a large cluster of overlapping circles. This does reveal the increased population areas, and the increased distribution of area codes in those areas, but does little to reveal the size, shape and nature of the area codes. Likewise, area codes in sparsely populated areas (such as 530, centered on Redding, CA), do not adequately demonstrate the full geographic area that the area code covers.
In hindsight, it would be preferable to perform a transformation on the shapes of the area codes and adjust their size according to population.
