The taiko drumming on trash cans at the opening ceremony was kind of cool, but being a true geek I have to say the cool thing of the day goes to Visual Analytics. By the way, ten brownie points and kudos to whoever had the bright idea of having the SAS demo room open on Sunday. Often, I get there for about five minutes between sessions. At one conference, I didn’t manage to make it at all. Having it open today before any papers was a great idea.
Today I only got there for about 15 minutes but it was the perfect amount of time to meet Don Chapman and hear about Visual Analytics which is a brand new product being rolled out by SAS and one of the things on my list to check out at SAS Global Forum. See picture of Don Chapman lest you need to identify him and get your own demo, which I highly recommend you do because this is way cooler than I am going to be able to describe it.
The first question I asked was how is this different from JMP. His answer was that while both were very graphically oriented, Visual Analytics is more so. Secondly, Visual Analytics runs on the web while JMP runs on your desktop. Third, VA (we’re now on a nickname basis we’re so tight) is good for running against really huge data sets.
What do we mean by more graphically oriented? Well, check this out.
If you want to explore the relationship further, you can click on any one of those boxes, say it is marketing expenses and revenue, and you’ll see on the right side of the screen a plot of the two variables.
This ability to just click and get more information is included throughout. For example, with the box plot, you can drag variables from the left pane to the large middle pane and get a box plot. If you want more detail about one of the variables click on the box and a pop-up window gives you basic statistics on that variable.
This is not, at present anyway, a very sophisticated statistical tool. What it IS, is a way to look at large amounts of data without knowing anything about SAS and being able to explore relationships in your data.
If you’re a really serious statistician, you might turn up your nose at things like the correlation matrix boxes.
Since I have never been a really serious anything, I think it is a great idea. In fact, I think they stole the idea. from me. For years, if we have had a project with a large number of correlations, I would sit down with a highlighter and mark those I expected to be substantially positive(negative) and were in one color, those I expected to be correlated but were not in a second color and those I did not expect to be correlated and were in a third color.Say the first color was blue. If I looked at the huge matrix and saw mostly blue and little of the other colors, that was good. (Good being a technical term used by not really serious statisticians to mean not sucking. )
This isn’t the exact same thing but it is the same concept, of taking a huge amount of data, say millions of records on 20 variables and being able to see at a glance which of those are the largest relationships.
It also does several other types of plots
As well as maps
It is full of coolness and brand spanking new. It was released in March and I don’t know anyone who has it installed, so I am the first kid on my block to see it. So, part of my purpose of SAS Global Forum – learning what is coming down the pike – has been fulfilled.
Go find Don in the demo room and check it out. As you can see, I have helpfully included a photo to assist you in Don-identification. Or you could just look for the big sign that says Visual Analytics.