SAS Visual Statistics Decision Tree Demo

Gain better insight on segments of your data through Decision Tree Analysis, used in both SAS Visual Statistic and SAS Visual Analytics.

REGISTER FOR A FREE TRIAL OF SAS VISUAL STATISTICS
SAS® Visual Statistics helps you explore data and quickly build and refine predictive models so you can zero in on the right answers to your biggest business questions.

Once you register, you can:

* Gain unlimited access to the technology.
* Start building and refining models on the fly.
* Find more value in your data.

http://www.sas.com/en_us/offers/14q3/visual-stats-try-before-you-buy.html


SUBSCRIBE TO THE SAS SOFTWARE YOUTUBE CHANNEL
http://www.youtube.com/subscription_center?add_user=sassoftware

ABOUT SAS
SAS is the leader in business analytics software and services, and the largest independent vendor in the business intelligence market. Through innovative solutions, SAS helps customers at more than 75,000 sites improve performance and deliver value by making better decisions faster. Since 1976 SAS has been giving customers around the world The Power to Know.®

VISIT SAS
http://www.sas.com

CONNECT WITH SAS
SAS ► http://www.sas.com
SAS Customer Support ► http://support.sas.com
SAS Communities ► http://communities.sas.com
Facebook ► https://www.facebook.com/SASsoftware
Twitter ► https://www.twitter.com/SASsoftware
LinkedIn ► http://www.linkedin.com/company/sas
Google+ ► https://plus.google.com/+sassoftware
Blogs ► http://blogs.sas.com
RSS ►http://www.sas.com/rss

---

Closed Caption:


Hi.
My name is Suneel Grover,
principal solutions architect
for SAS.
And today, I'll be
sharing a demonstration
using Visual Statistics to
showcase decision tree analysis
for a variety of predictive
marketing applications.
With that said,
let's jump into it.
In this demonstration,
we'll be performing
decision tree analysis with
Visual Analytics and Visual
Statistics.
The business use case will be
to analyze clickstream data
for the purposes of performing
a predictive marketing analysis.
What we're looking at in
the middle of the screen
here is the number
of visitors who
came to sas.com and with
respect to a business goal,
in this case a conversion event.
Approximately 90% of those
that visited did not convert
and about 10% did.
The entire point
of this analysis
is to identify attributes
that uniquely describe
conversion-like behavior and
differentiate those segments
from those that do
not convert according
to very valuable insights
for marketing teams.
So to perform this analysis,
it's actually quite easy.
We just come over here and refer
to our attributes over here
to the left.
And I'm going to
just begin starting
to select certain
attributes that I
want to add into my analysis.
First and foremost,
what were people
interested in when they
came and visited sas.com?
And I'm going to
select those attributes
and drop them into the analysis.
As you will see, the software
will quickly churn and identify
which attributes are important
and which attributes are not
important with the
black and gray font.
Now, before we start
breaking this down,
let's add a little
more information,
because as analysts,
we're curious.
And with this type of
speed, we can be as curious
as we want to be.
So next, I want to look at where
were people originating from,
or what traffic or
marketing channels did
they come from before
they landed on our site?
And I'm going to
grab those attributes
and also add them
into the analysis.
The tree continues to grow.
And lastly, I want
to understand how
engaged were these visitors?
How long were they
spending on our site
when they came and visited?
How much content
were they consuming?
And at this point,
I actually want
to now sit back and take
a look and interpret
what we've created.
Now up here at the top,
just as I mentioned earlier,
we had about a 90-10 split
in our marketable population
that came to the site.
And the first attribute that was
most important in partitioning
those who were converting and
those who weren't converting
was visitor engagement.
And so those who branched
right who had a score of six
or higher for engagement landed
in this second level node.
And we notice the
proportion of conversion
and non-conversion
behavior shifts from 90-10
to a 65-35 split.
We are seeing
healthy improvement
in identifying signal related
to conversion behavior.
Now, interacting
with engagement,
we find that those who haven't
engaged scored higher than 6
as well as originated
from organic search, they
arrive here.
And this is our first
very attractive segment
in this analysis.
And what we can
see he is about 73%
are converting when they meet
these two pieces of criteria.
Now, the color
shifts because we now
have a higher proportion
of converters,
and that's what the
green indicates.
As you can probably
tell now, we'd
like to know the story behind
this particular segment,
or this particular segment.
The wonderful thing
about the tree
is it churned through
all this different data,
it prioritized which
attributes were important,
and then identified
within the distribution
of those attributes
what cuts optimized
how to get to the identification
of those attractive segments.
Now, with that said, what do
we do with this information?
How do we take action on it?
There are a couple
things we can do.
First, we can hit this
little drop down arrow.
And we can first and foremost
export the score code.
So if we want to take the
intelligence of this tree
and operationalize on
it or productionalize
its intelligence in
a marketing workflow,
say a marketing
automation system,
all the code is
pre-generated for us
and we can now deliver that
to our marketing peers.
A second option
that we can utilize
is that we can select
a segment, right click,
and select Create
Visualization from Node.
And what this will do is
it will open a new window.
And with the intelligence
to identify that segment,
it will be auto
coded into my filter.
And then I can just
select attributes here
from the left to create
an audience segment table.
Then now my marketing campaign
peers can utilize the target.
Now let's leave the
world of Visual Analytics
and turn on some capabilities
from Visual Statistics.
I'm going to come over
here to the Properties
and show for those of you that
like to build models and like
to tinker all the available
options for you to leverage.
You'll notice there
are different things we
can do to customize a tree and
potentially further optimize
to our exploratory analysis.
We also can open up
diagnostic plots.
By turning on these
assessment plots,
we can assess how healthy
of a model we're building.
For example, we can
utilize assessments,
such as a lift chart,
an ROC plot, or even
a misclassification
chart to understand
how many observations
we're correctly predicting
and how many we're
incorrectly predicting.

Once we evaluated the different
assessments of our modeling
exercise and we become
satisfied with the model
that we want to select
and share those insights
with our marketing
peers, I just want
you to think about how you
might be able to exploit
this information.
You could have
information that could
support digital personalization
on a website or an app.
It also could support some
sort of marketing campaign
from an outbound perspective,
say an email campaign,
and you want to target
these segments that we
identified as attractive.
There's a number
of different ways
to utilize this
information, and we
hope you found the
demo today compelling.
So that was the demonstration
of Visual Statistics.
For predictive
marketing applications
using decision tree
analysis, we really
appreciate your time today.
We hope you enjoyed it.