"It's
exciting to see how our staff research bears fruit at unexpected times
... Who knows what will crop up next?"
Recently Azavea went through another round of office expansion, almost doubling our office size. We knocked down walls, carved up new conference rooms, added a bike garage (as opposed to a bike tree), and more. We now have lots of new space, and quite a few new people. One of the questions that simmered while we watched the work complete was: where are we going to sit? Our staff is full of busy, smart, sophisticated people who can’t be bothered to do their own spatial analysis. Can’t we come up with some way to take the thinking out of this equation? In addition, this question is inherently spatial, so it sounded like a great opportunity to leverage our spatial research.
Map of Azavea’s office showing an employee’s ideal desk location based on entering weighted preferences in DecisionTree.
The basic premise is that when individuals moved their desk, they will move toward something they desire, and away from something they don’t. If you are allergic to printer toner, you don’t want to sit next to the printers, and if you really like the sun, you definitely want to sit next to the windows. The ultimate location of an employee’s desk takes into account all sorts of factors, and comes to a solution that is often unique to the individual. Does this sound familiar? Indeed! Managing these types of decision factors is the basis for Azavea’s DecisionTree® framework.
Using these principles, it became apparent that software developer, David Zwarg’s research was well suited to address this problem. One of David’s ongoing research projects at Azavea is collaborating with Dana Tomlin at the University of Pennsylvania to develop an advanced raster cost-distance algorithm. The innovation behind this raster cost-distance algorithm is a wave propagation model, which is not constrained to the grid imposed on the raster data. Bonus!

To start, David picked some key landmarks in the new office, and generated a cost-distance raster for each of them. His list of raster datasets generated include: cost-distance to the refrigerator, cost-distance to the bike garage, cost-distance to the printers, cost-distance to the windows, and more. In all, there were 14 layers — or decision factors — that David was able to incorporate, based on the new office floor plan.
Next, he converted the raster datasets to the Azavea Raster Grid (ARG) format. What is this format, and why convert data from raster grids? ARG is a grid format that we use internally (not to be confused with “Argh!”, which is also used internally) and has been optimized for fast processing and storage speed, in addition to being the format used by DecisionTree.
Finally, David plugged the raster datasets into a demo DecisionTree application, and published the application on the Azavea Intranet a couple weeks prior to the completion of the office expansion. The application contains a base map that is the architectural floor plan of the new office space. Azavea staff members could now use DecisionTree to locate the places in the office that suited their preferences. Adjust a few sliders, click update, and the application shows the best place in the office, based on your criteria! No more guesswork required.
It’s exciting to see how our staff research bears fruit at unexpected times. Across the gamut, from Open Source projects to geoprocessing to pro-bono cartography, our staff research brings a wealth of experience to their work (and play) – who knows what will crop up next?