Visualizing the Polyglots

I've had a lot of conversations lately regarding the adoption of polyglot programming and persistence in applications and database solutions. And if not actual adoption, at least moving toward adoption of the concepts and approach of a polyglot world. I get excited about our solutions harnessing the best features of available technologies. 

Even in this exciting time where developers, analysts, engineers, and scientists have the ability to pick and choose the languages, tools, processes, and methodologies available, I see frequent instances of folks over-advocating for their tool of choice. I'm sure I've been guilty of this myself, especially earlier in my career. Certainly some of that is a result of the reality of using what's available to the user. I'm sure many people can point to several unnatural "tools" that have been created in Excel (with or without the enabling assistance of VBA) that probably would have been better created in...well, anything else, really. To be efficient and effective, we have to be more aware of what's out there and have a variety of tools available.  

In the realm data visualization, especially, I'm convinced of the need to have a suite of tools available. When I am faced with a relatively unknown data set to explore, I'm typically working through a very different workflow than when building out the requirements for a static report that contains a few bar charts and will be distributed throughout our organization. Here is my quick summary of my preferred tools for different types of visualization work:

Data Exploration

Tableau. Hands-down. I can hit cubes, databases, spreadsheets, text files, etc., from the same interface. For me, the comparison is not even close when compared against the likes of R for data exploration, let alone visual data exploration. Yes, Tableau ain't cheap. Were I constrained differently on the time/money continuum, maybe I'd choose something else. I doubt it. 


Yep, I went there. The "D" word. Here's my definition of a dashboard, regardless of how its defined elsewhere: 

  • Loads in any browser. Quickly.
  • No extra clicks, exports, or parameters for advanced functionality.
  • For the love of all that's good in this world, no pie charts in default views.
  • I want to be able to change what I care about from time to time, without any required development time.

For our dashboard needs, I'm choosing Domo. I'm a big fan of what they're doing, and, more importantly, I'm a big fan of what using Domo has done in our organization.  

Statistical Analysis

Given previous posts, it's no surprise that I very much prefer JMP. Histograms, various distributions, the interactive functionality within the charts and the data set are all fantastic. On a project using SAS in the workflow, I prefer to use Enterprise Guide to build the program but connect JMP to the data sets of interest. 

Tibco Spotfire intrigues me in this space, too, especially for visualizations like box plots, clustering and network analysis that Tableau doesn't do.  (Yes, I know I can create a box plot in Tableau with some calculated fields, etc. I can also just use a tool that already does box plots.)

Prototypes and Report Mockups

This one's a toss-up between Tableau and Excel. Probably more than half of the time I use Tableau for a quick report mock-up. If there are dates involved that probably moves closer to 100%. 

I do like using Excel for mock-ups of data tables and visualizations based on several different calculations, and Power View had made this even easier. For a mock-up, typing "=<formula>" is just quicker than going through the admittedly still-simple process to add a new calculation in Tableau. 

Enterprise Platform

For out-of-the-box functionality, Reporting Services wins for me right out of the gate. The fact that we already license SQL Server is definitely a consideration. I also weigh the ease with which I can find online resources and developers with SSRS experience. The charts and graphs are mostly acceptable, and the folks in Redmond have made some nice attempts to keep up with the market.

I like to call SSRS the "house brand" reporting platform. Just like the house brand pancake syrup from Target really doesn't compare to, say, real maple syrup or even Mrs. Butterworth's original, and any serious breakfast eater will dismiss the house brand, my kids like the Target brand just fine. All the Target brand label is missing to complete this analogy is to say "Exports to Excel!"

I lament that we haven't yet brought the "pure maple syrup" to our Enterprise Reporting Platform. We are exploring several options, each of which provides rich visualizations that exceed the capabilities of SSRS. 

Wrapping up...

I'm very excited to see where we head this year as we expand our visualization and reporting efforts into the cloud. Vendors are making it easy to spin up pre-configured machine images, ready to visualize our data or connect to huge repositories either already in the cloud or still on-premises. 

After writing down my thoughts on this topic, I pulled out Now You See It, by Stephen Few. He uses the term data visualization "to cover all types of visual representations that support the exploration, examination, and communication of data." We use data visualizations for many different purposes in widely disparate disciplines on multitudes of different data sources. I think it only makes sense that we should also have a variety of tools available, too.