For a while now I’ve been very interested in data visualization — specifically how visualization can be used to gain understanding of really large data sets. Actually that’s not true. Like most people I’ve always been drawn to data visualization, if for no other reason that we humans are really quite bad at gaining any understanding from large complex data sets when they are only presented numerically. All but the most glaring patterns are lost to us once we are faced with more than a just a few columns and rows of data.
A few weeks ago a very interesting presentation by Tableau Software at W&M’s Mason School of Business covered some of the reasons why visualization is such a powerful tool, and specifically discussed how computer visualization allows us to exploit our highly developed visual perception/cognitive processes to identify patterns in large data sets. A couple of similar presentations from Tableau covering some of the basics of data visualization and human perceptions are covered here (human perception and design).
Geospatial Data Visualization — What’s Out There?
Data with a geospatial component is of particular interest to me. For a while now we have been able to “see” and review this type of data using systems such as Google Earth. However, thus far we’ve only been able to display fairly static data and then with only limited user control — typically tilt, pan, zoom, but with no control of which specific subsets of data are displayed. Furthermore the data that we can see is often fairly dated.
My hope has been that (ideally) newer systems would allow us to visualize even potentially larger data sets, allow greater control of which subsets of data are displayed as well as how they are displayed, and include near real-time data.
For the above reasons, I could not have been more pleased when a few months ago I came across the absolutely amazing ‘Earth” Project by Cameron Beccario. “Earth” was based in part on the similar but much less ambitious “Wind Map” Project by Fernanda Viegas and Martin Watterberg. Both of these meteorological visualization projects are similar in that they tap into the huge amounts of free data (actually incredibly detailed models based on a huge number of observations) available from NOAA and/or other online resources, and run largely on open source code (details of data sources as well as code are available on this page). While “Wind Map” provided a revolutionary way of visualizing near real-time wind data for the United States, “Earth” significantly expands on this idea by allowing the user to see world-wide meteorological data, including wind patterns and temperature (by elevation), ocean currents, humidity, air density, mean sea level air pressure, etc., in various user-controlled combinations.
Check out “Earth,” embedded below:
The data available for visualization includes a few months of historical data, current conditions, as well as a five-day forecast of predicted conditions. Additionally, the user can not only zoom in and out, but also (wow!) change the globes projection type and central point, which is something that I had never seen done before on the Web. The number of available features, types of data and projection options are truly amazing and are so numerous as to make it unfeasible to cover here — I encourage you to do a bit of exploration (most controls are accessed by clicking on “earth” in the lower left corner).
As any good computer visualization should do, you not only get a the big picture in a visually stimulating and perceptually easy to interpret way, but you can still get at specific numerical data, all you need to do is simply click on any location on the world and you have access to the specific conditions at that location (see below).
Wave of the Future?
Free online projects such as “Earth” that allow us to visualize increasingly available large data sets of various types are part of a growing trend. For now, as far as I have been able to find, “Earth” is still the most sophisticated of these projects (at least for my interests and in term of user control), but hopefully we’ll be seeing some real challengers in the near future.
In the world of social media, a large number of “heatmaps” (such as this example) display gathered information on things such as the source and location of Twitter posts and of other social media applications. Another interesting example is called Facebook Friend Wheel which allows you to generate a visual representation of your social network — using this tool and your brains ability to process visual information you may quickly find some very interesting relationships among your friends that you may not have otherwise realized existed.
In the area of public health, Google provides an estimate of flu trends for the United States as well as several other countries based on search terms. Another Google Chrome sponsored project helps provides a very illustrative and compelling view on how the trade in small arms and ammunition is distributed world-wide.
Interestingly, in many cases the owners/collectors of the data being displayed are not the ones developing the tools to visually analyze their own data. One example of where this is not the case is the US Geological Survey’s excellent new earthquake mapping page that allows users to create a seismic map of any region of the world based on a large number of criteria and potentially over 100 years of earthquake data. This website also allows the user to download a list of these events as well in a tabular form. A similar ongoing effort for the US Census Bureau does not seem to be as successful, but is at least a step in the right direction.
My hope is that these types of projects will thrive in the future. Of course this can only happen if large data sets are made available for these efforts to “mine” and display. The increasing commoditization of data could well counterbalance growth of these data visualization projects, so lets hope that us much as possible the owners of these data sets continue to share.