I recall this day in school where the professor was drilling into our heads "select … from … where … Anyone? Anyone? Anyone seen this before?" Oh wait, that was somebody else's schooling. As we were running our database drills I leaned over to my friend and said, "WHO in their right mind would do this for a living?"
Turns out I am, or at least was, that person.
I love normalization and putting messy data into neat rows and columns. When all your data is inside a single database (or multiple on a shared server, or multiple servers that are linked) it's easier to find errant data. Even with all the limitations of SQL http://en.wikipedia.org/wiki/SQL. And by SQL, I mean the old school T-SQL standard. The new reserved words in MS are awesome and powerful but I leave that functionality to reporting tools. Please don't pivot my raw data.
What do you do when the data is not in a homogenous source? And you have data that can't be loaded into a RDBMS? And you hypothetically lost some block groups?
You use Alteryx.
In my source data I have a combination of ZIPs and block groups with their associated polygons and centroids. This data resides in .csv and .tab files. Using Alteryx's join tool I can quickly locate the block groups that are in one file and not another. Using formulas I can make the necessary computations and load the missing block groups into the other file. My block group sources are in-sync; a quick verification with another join tool checking the right and left outputs are empty.
Now I use SpatialMatch to look for block groups outside the ZIPs. This is what a co-worker calls a "retinally weighted" approach – I was spot checking that my block group centroids were falling inside the ZIP polygons.
No way would a database have supported this. Even a spatially aware one. And not at these speeds. My geography files include the entire US and Puerto Rico.
So, again, Alteryx has made ad-hoc data analysis easy to do.
