This month Tim Costello (blog | twitter) gave a presentation on Rapid Fire Business Intelligence using Tableau.  Tim works for InterWorks, where he specializes in ETL and Data Analytics.  When you visit his profile and look at his picture, he wants you to keep in mind he was participating in No-Shave November at the time.


Tableau is not SQL Server Reporting Services, it is an Analytics Tool.  The strength of SSRS lies in tabular reports, using string data.  You use it to answer a question that you already know.  Contrast that with Tableau, whose strength is in helping you identify a question you did not know about, but should be asking.

The Big Picture

Most things are done at the aggregate level to enable you to see the big picture visually.  Machines are able to recognize patterns in text very easily, but for us mere humans, it is easier to do this visually.  This is what Tableau focuses on doing.  Furthermore, Tableau hired a professional color designer, who designed some custom color palettes for users who are colorblind.  You can see a nice example of this that uses good ole AdventureWorks for the data source.

The internals of how you setup Tableau are similar to SQL Server Analysis Services.  You will see all the familiar dimension and fact tables.  However, the setup is more visual and user friendly.  It is intended for users, well…power users, instead of designers and engineers.  Tableau is cross platform and you can point it to SQL Server, SSAS, Oracle, Firebird, MySQL, or many other data sources.  Similar to Adobe Reader, Tableau is able to use disconnected data so that you may view and analyze it offline.

After you have connected to your data source, you are presented with a nice, easy to use WYSIWYG designer to setup your Data Analysis.  At this point, it is helpful to run through the Categorical Analysis.  This is a basic statistical breakdown of your data; how many rows, how are your values spread out, etc.  When setting up your design, consider one of the Principles of Data Visualization, “don’t overwhelm your customers with data.”  You can embed SQL queries directly into the connections, however the best practice to create views on the database server so that you can reuse code.

As you begin to analyze your data, it pays to be cognizant of how standard deviation plays into your data.  Seventy-five percent of your population should be contained within one standard deviation, and ninety-five percent of your population should be contained within two standard deviations.  So you want to keep this in mind while looking for outliers in your data set.

While checking out his example, we did see one such outlier; Null Island.  Null Island is conveniently located at the intersection of the Prime Meridian and the Equator.  For my next vacation, I think I may go visit.

Visual Resume

There is a public version of Tableau available.  When using this version, you are limited to 500,000 rows and your data must be uploaded to their servers.  Obviously, this is not a viable option for most production environments.  However, this could serve well for a proof of concept or test environment.  Tim has a demo of this where he has visual version of his resume posted.  By leveraging the visual tools in Tableau, you are able to see a graphic breakdown of Tim’s skill set and experience.  Not that he’s looking…  😉

In conclusion, please keep in mind that analytics is not reporting, and reporting is not analytics

User Group Business

We are trying to rearrange the San Antonio SQL Server Users Group to meet on Mondays.  Doing so, would bring us closer to the goal of having each major metropolitan area hold their respective meeting on a different day of the week.  Having the days aligned in this manner will facilitate having out of state or “big name” speakers make the Texas Tour.

Austin is looking to experiment with adding a second SQL Server Users Group that is located in the downtown area.  This will compliment the existing group which meets in North Austin.

And finally, the news that you have all been waiting for…

We are getting closer to hosting a SQL Saturday in Austin.  The tentative date is September 10th, so mark your calendars.  Let’s all think happy thoughts to help make this a reality.