Feeds:
Posts
Comments

Posts Tagged ‘Visualization’

Visualization of data in analytics is getting to be a very interesting space to explore today. Traditionally viewed as an efficient way of transferring a large amounts of information quickly, fairly common delivery methods can be seen in a variety of dashboard examples found at the Digital Insight site gallery. What is more interesting is an emerging trend to begin leveraging visualization techniques to become even more powerful when multiple people can access them for collaborative reasoning and shared insight. By building graphical displays that promote the exchange of ideas and insights, we can begin to explore the value of information in new ways. Our good example of this is the living laboratory for exploring these concepts known as Many Eyes.

As part of IBM’s Collaborative User Experience research group, the ManyEyes project  hosted at alphaworks.ibm.com represents a truly fascinating idea where people can develop visualizations in order to see and exchange information in novel ways. In 2007, Fernanda Viegas and Martin Wattenberg created Many Eyes, a web site where all us can upload data, create interactive visualizations, and carry on conversations. Their design goal is to transform visualization from a solitary activity into a collaborative one.

Visualization options available in Many Eyes (as of April 2009) range from the ordinary to the experimental to include:

  • Relationships among data points
    (Scatterplot,  Network Diagram,  Matrix Chart)
  • Comparing a set of values
    (Bar Chart,  Block Histogram,  Bubble Chart)
  • Track rises and falls over time
    (Line,  Stack, and Stack Graph for Categories)
  • Parts of a whole
    (Pie Chart,  Treemap,  Treemap for Comparisons)
  • Analyze text
    (Word Tree,  Wordle,  Tag Cloud,  Phrase Net)
  • Maps
    (US County Map,  World Map)

All the visualizations in Many Eyes accept the same data format (text analysis such as tag clouds also understand free text) where each data table consists of rows where the values in each row are delimited (separated) by tabs. The first row of the table should be “headers” that describe the columns. This format can be easily exported from databases or spreadsheets. They currently impose a size limit of 5 MB per file. This format is well documented and ManyEyes provides enough examples to get you started.

For example consider the spreadsheet data we have all come to love and hate at budget time (vastly simplified here for our purposes) typically organized and summarized as follows:

budgetexampledata

Simply uploading this into ManyEyes now quickly gives us an interactive tree map hierarchy where the data becomes a little more useful and values begin to jump off the page as follows:

budgetexample1 

The same data can be expressed as a matrix chart  summarizing the multidimensional data set in a grid. A matrix chart can be seen as the visual equivalent of a “cross-tab” or pivot table.

 

matrixchart

 

One of the more interesting visualizations is what we can do with text.  Aside from the more common Wordle (the size of a word in the visualization is proportional to the number of times the word appears in the input text) and Tag Clouds this site also offers Phrase Net and Word Tree visualizations.  For example see the United States Declaration of Independence as a wordle (prepared from the http://www.wordle.net/ site).

declarationindependence1

 

And the same document rendered as a phrase net.

 declarationindependence_phrasenet

 

A word tree is a visual search tool for unstructured text. Selecting a word or phrase and will display all the different contexts in which this phrase appears in the text. The contexts are arranged in a tree-like branching structure to reveal recurrent themes and phrases. The image below is a word tree made from the same Declaration of Independence. Font sizes show frequency of use, so you can see that among many uses of “we,” in this example the most frequent context is the phrase “hold” and “have” expressed as follows:

 declarationindependence_wordtree
 

Many Eyes is a site I recommend (although you may end up spending hours there<g>) as a starting point for you to begin thinking about visualizing data in new and meaningful ways. At the very least you will come to appreciate the value in having this living lab readily available to discover for yourself that finding the right way view your data is as much an art as a science.

Read Full Post »