Analysis of LendingClub.com data self-reports of employment tenure from 2007 to 2011.
In this project, I downloaded two public data sets from LendingClub.com. One data set contains rejections from 2007 to 2012, and the other contains all loan data from 2007 to 2011, the early years of LendingClub. Within those data, there is a self-reported employment tenure for each data point. I parsed all the data (800,000 points) by state, and then looked at the average tenure for each state. The options for tenure were limited, with the largest option being "10+ years" and the smallest option "<1 year". For these computations, I set "<1" to be .5 years, and I did two averages, one for exactly 10 years, and one for 20 years. The two results were identical in many places, and only differed by a few days in most days; I used the average of the two values. This was all done in Python using pandas.
I used d3 and created two cartograms. One image uses all the data, and the other image excludes 7 outlying states. The 7 outliers had very few data points. These images are in the form of the two html files.