[Webinar] SpiraTest - Setting New Standards in Quality Assurance
How to effectively realize data visualization
1. CONCLUSIONS PAPER
How to Effectively Realize Data Visualization
Gaining Insights from Your Data Faster than Ever and Sharing the Insights with Others
Featuring:
Justin Choy, Global Product Management, Business Intelligence,
SAS
Jennifer Marchi, Global Product Marketing, Business Intelligence,
SAS
Insights from a Webinar in the Applying Business Analytics Webinar Series
3. 1
How to Effectively Realize Data Visualization
Quickly Visualize Big Data Analysis on the iPad®
and the Web
Organizations are inundated with data – terabytes and petabytes of it. Data pours in
from every conceivable direction: from operational and transactional systems, from
scanning and facilities management systems, from inbound and outbound customer
contact points, from mobile media and the Web. It is structured and unstructured; it is
messy and overwhelming; it flows at greater speed than ever.
The optimist’s vision for this tidal wave of data is that organizations will be able to harvest
and harness every relevant byte of it to make supremely informed decisions.
However, traditional architectures and infrastructures are not up to the challenge. IT
teams are burdened with ever-growing requests for data, ad hoc analyses and one-
off reports. Decision makers become frustrated because it takes hours or days to
get answers to questions, if at all. More users are expecting self-service access to
information in a form they can easily understand and share with others.
“Many challenges arise when working with this much data,” said Justin Choy, Global
Product Manager for Business Intelligence Solutions at SAS. Choy joined Jennifer
Marchi of SAS Business Intelligence Product Marketing for a webinar in the SAS
Applying Business Analytics Series. “One problem is that it takes a long time to
aggregate all that data. To get around this problem, IT has to create summaries of the
data, or a cube with predefined hierarchies for better performance.”
“Once you have a summary or a cube of the data, you can look at it more quickly,” said
Marchi. “But what happens when you realize you need to look at the data in a different
way than the summary or cube provides? You would typically have to turn to a data
administrator or someone else in IT, and request another view of the data. In most
organizations, accessing this new view will take some time, sometimes even a day,
which, in turn, delays your analysis.”
Another problem is that some business questions are best answered by using all
available and relevant data, which might be too much data to process within the
decision window.
Here is where high-performance computing and data visualization techniques come in.
“Data visualization is shaking up the world of analytics, changing the way you work with
data,” said Marchi. “You’ll find insights from your data faster than ever and be able to
share those insights with others.”
Learn More
View the on-demand recording
of this webinar:
sas.com/reg/web/corp/1838098
Other events in theApplying
BusinessAnalyticsWebinar Series:
sas.com/ABAWS
For more about SAS®
Visual
Analytics:
sas.com/visualanalytics
To see a video demo of
SAS®
VisualAnalytics:
youtube.com/
watch?v=1DUM4YdnzzA
Related white papers:
Big Data: Harnessing a
Game-Changing Asset
sas.com/reg/gen/corp/1583148
In-Memory Analysis: Delivering
Insights at the Speed of Thought
sas.com/reg/wp/corp/41962
High-Performance Analytics
Insights: Big Data, Bigger
Opportunities
sas.com/reg/gen/corp/hpa-insights
4. 2
SAS Conclusions Paper
Introducing SAS®
Visual Analytics
“SAS® Visual Analytics is a high-performance, in-memory solution for exploring massive
amounts of data very quickly, using SAS’ industry-leading analytics,” said Marchi. “It
enables users to spot patterns, identify opportunities for further analysis and convey
visual results via Web reports or an iPad®
.”
With SAS Visual Analytics, business users can explore data visually on their own, but
it goes well beyond traditional query and reporting. Running on low-cost, industry-
standard blade servers using Hadoop distributed processing, it delivers answers in
seconds or minutes instead of hours or days. This kind of processing speed with
massive data sets makes it possible to find important insights quickly – insights that
were not evident before.
For example, marketing campaign optimizations that previously took eight to 10 hours
or more can now run in less than three minutes. Bank risk calculations that once took 18
hours can now be completed in 15 minutes.
High-performance computing completely redefines what’s possible. Analysts can use
complete data sets instead of sample subsets. They can apply sophisticated analytic
techniques, not just simple calculations, to generate descriptive and predictive insights
from all that data. And they can put this power into the hands of a broad range of users
through a dynamic, visual interface.
Consider some possibilities. Retail marketing analysts are always looking for ways to
optimize the offers and promotions they make to customers. When they can include
more data from online sales and hundreds of stores – plus demographic information,
purchased market data, social media data and any other relevant data source – they
can determine “next best offer” recommendations with greater accuracy and certainty.
They are not limited to a subset of customers; they can get on-target recommendations
for entire populations, quickly and easily.
Consider fraud detection models for banking and finance. With SAS Visual Analytics
embedded in the analytics life cycle, analysts can look at more data – more often and
more quickly – for new insights that lead to better models that can be quickly adapted to
address new and emerging patterns of fraud.
Retail, telecommunications and financial services are showing the greatest early interest
in SAS Visual Analytics because of its ability to analyze large customer and prospect
pools at an individual level while easily managing a collaborative analytics environment.
In time, data visualization capabilities will be extended across the SAS software portfolio.
SAS Visual Analytics combines
an easy-to-use dynamic
interface and in-memory
processing to enable all types
of users to visually explore
big data, execute analytic
correlations on billions of
rows of data in minutes or
seconds, understand what the
data really means, and deliver
the results quickly wherever
needed via Web reports and
mobile devices.
Do you simply want to collect
large amounts of data, or do you
want to understand and take
advantage of its full business
value? What if you could easily
access more than a billion
rows of data, and then very
quickly explore all of that data
using visualization techniques
combined with powerful
analytics? You can.
5. 3
How to Effectively Realize Data Visualization
The Six Components of SAS
®
Visual Analytics
SAS Visual Analytics includes the following components.
• The SAS®
LASR™
Analytic Server quickly reads data into memory for fast
processing and data visualization, with direct communication to clients.
• SAS Visual Analytics Explorer is a discovery and visualization tool for ad hoc data
exploration and analysis.
• SAS Visual Analytics Designer provides a graphical, drag-and-drop environment
to create standard and custom reports and dashboards.
• SAS Visual Analytics Mobile is a tool for viewing reports by connecting to servers
and downloading information on the go.
• SAS Visual Analytics Hub provides a central location to launch the various
elements of SAS Visual Analytics.
• SAS Visual Analytics Environment Administration allows administrators to
manage users, security and data.
SAS Visual Analytics is tremendously powerful for its in-memory and massive parallel
processing capabilities, but it is still cost-effective, said Choy. “One of the ways we’ve
made it cost-effective is to use commodity hardware, taking advantage of Hadoop
for data persistence on blade servers from HP and Dell, as well as dedicated data
appliances from our partners Teradata and EMC Greenplum.”
SAS LASR Analytic Server has been tested on billions of rows of data and is extremely
scalable, bypassing the known column limitations of many relational database
management systems (RDBMS). This processing power is coupled with SAS Analytics,
which differs from other solutions that simply move data from a SQL database into
memory. Other solutions cannot support regressions or logistics models because such
capabilities are not built into databases.
“The speed of in-memory
architecture offers tremendous
benefit. Organizations can
explore huge data volumes
and get answers to critical
questions in near-real time. SAS
Visual Analytics offers a double
bonus: the speed of in-memory
analytics plus self-service
eliminates the traditional wait
for IT-generated reports.”
Dan Vesset
Program Vice President of Business
Analytics Research, IDC
6. 4
SAS Conclusions Paper
Fast and Easy Visual Data Exploration
“One of the first things a data analyst wants to do when presented with a new set of
data is to get some idea what it’s about,” said Jim Goodnight, CEO of SAS. “What are
the outliers? What variables are related to each other? Up until now, analysts would have
to spend hours and hours doing samples so they could actually get jobs to run. With
SAS Visual Analytics, you can cruise a billion records to create graphs and plots one
after the other. It really allows you to understand your data better before you go into the
high-performance analytics.”
Marchi and Choy demonstrated how SAS Visual Analytics Explorer can conceptualize,
visualize and investigate the data in different ways to find the best answer a particular
business question. To see SAS Visual Analytics in action, check out the four-minute
demo at youtube.com/watch?v=1DUM4YdnzzA.
Visually Explore the Data
“By dragging and dropping hierarchies or categories onto the visualization pane, I can
quickly and easily see how my revenue expenses are doing by region,” said Choy,
demonstrating the system. “I can hover over the bubbles and see that my West region
is doing better than other regions. I can double-click on the bubble and go into that
region for more detail. It’s as simple as that, and all of the data being presented to me
is based on a billion rows of geographic, product and financial information for
this manufacturing company.”
Figure 1: The interface uses intuitive, drag-and-drop actions and visualizations.
Users can explore all data,
execute analytic correlations
on billions of rows of data in
seconds, and visually present
results to the Web and mobile
devices. This kind of processing
speed with massive data sets
makes it possible to quickly
identify patterns, trends and
relationships in data that were
not evident before.
7. 5
How to Effectively Realize Data Visualization
Create Your Own Hierarchies
If you need the data organized a certain way for the business problem at hand, you
don’t have to go to your company’s data specialists to request a new view of the data.
“With SAS Visual Analytics, you have access to all of the data, and based on all of the
data, you can create and modify your own hierarchies to explore the data, however and
whenever you want, without having to rely on someone else,” said Choy.
“It is so easy to do. I can create a new visualization and then go into the ‘Create a New
Hierarchy’ area, and then just double-click or drag and drop [from a menu selection box]
the items on which I want to build a hierarchy. Once I’m satisfied with that, I can give it a
name or just hit ‘OK.’ It’s that easy.”
Choy demonstrated how the new hierarchy appears in the visualization. Double-click on
a bar in a bar chart or on a slice of a pie chart, for instance, and you can drill down to a
different level or go back up a level. All of these views are changing within seconds, yet
drawing on a billion rows of data behind the scenes.
Choose the Best Visualization for the Data
“When you’re working with a billion rows of data, it is really difficult to quickly
understand what’s happening in the data simply by looking at rows and columns,”
said Marchi. “Even guessing what type of chart to use can be challenging. SAS
Visual Analytics simplifies all that by using intelligent autocharting to recommend an
appropriate type of visual for the information you’re trying to view. It could be a line or
bar chart, a scatter plot or box plot, heat maps or bubble charts – anything from the
portfolio of available visualizations.”
In the example shown in the demo, SAS Visual Analytics recognized that the data had
geographic elements – revenue and expenses by geographic regions – and offered the
data as a bubble plot superimposed on a map of the regions. In another example, Choy
showed how SAS Visual Analytics chose a line chart as the best visual for a selection of
two variables, then changed to a clustered bar chart when a third variable was added.
Of course, users can always select the desired of visual – anything acceptable to the
measures and categories chosen.
And remember, this speed-of-sight display is being calculated with 1.1 billion rows
and 47 columns of data.
Build-it-yourself hierarchies
eliminate the need for data
analysts to constantly seek help
from IT. It is easy to drill up and
down through the hierarchies to
slice and dice data on any level.
Autocharting capabilities help
users create the best possible
visualizations, including those
specifically designed to display
big data.
8. 6
SAS Conclusions Paper
Refine the View
What if you only want to view data for a certain region, product line or some other
variable? “SAS Visual Analytics has filtering capabilities that make it easy to refine the
information you’re viewing,” said Marchi. “Simply add a measure to the filter pane or
select one that’s already there, and then select or deselect the items on which to filter.”
What if the filter isn’t meaningful, or it skews the data in undesirable ways? A histogram
on the right of the screen provides a visual cue about the distribution of the data,
showing how the data is affected by filtering a particular measure, said Marchi, who
used expenses as an example. “The histogram shows that the majority of the data is
on the left, so if I filter from the right, it will have a minimal impact on the information
I’m viewing in the visualization pane. However, if I filter from the left, it will have a more
significant effect on what I’m viewing, because the majority of the data is distributed in
this area. Histograms allow you to save time by giving you an idea what effect the filter
will have on the data before you apply it. Rather than relying on trial and error or gut feel,
you can use the histogram to help you decide what areas to focus on.”
Understand the Distribution of the Data
Understanding the distribution of the data is now as simple as creating a new visual
and dragging the measure or category onto the visualization pane.
“By default, the system automatically calculates the best distribution for the data, but
the user can go under the ‘Properties’ tab and change that,” said Choy. “A histogram
shows how the data is distributed, but a better visual for this purpose is a box plot
[which graphically depicts numerical data through summary calculations such as
minimum, maximum, upper and lower quartile, and median]. I can create a new box plot
by creating a new visualization, selecting that visual, and then dragging and dropping
the visual on here. You can hover over the box plot and not only see the distribution of
the data, but also where the important 50 percent of your data is, as well as some basic
statistics about the data, such as average, median, standard deviation, minimum and
maximum, average standard deviation, whisker and quartiles information, and count.
“I can even add a new measure onto here, and we can do a comparison over two
different measures in the box plot. I can then add another category in here, such as
product line, for example, and now instead of doing 10 calculations on the fly, we’re
actually doing 40 calculations over a billion rows. And we’re getting the results extremely
fast, especially considering that the median, the middle value of the data, requires
scanning all 1.1 billion rows of data.”
9. 7
How to Effectively Realize Data Visualization
Figure 2: Interactive box plots help users quickly understand the distribution of the data.
Identify Relationships Among Variables
SAS Visual Analytics makes it easy to assess relationships among variables. Just
select a group of variables and drop them into the visualization pane; the intelligent
autocharting function displays a color-coded correlation matrix that quickly identifies
strong and weak relationships between the variables. Darker colored boxes indicate
a stronger correlation, lighter boxes a weaker correlation. Hover over a box to see a
summary of the relationship. Double-click on a box in the matrix for further detail.
“Previously, this type of calculation would have taken many hours, but now it can be
done in seconds,” said Choy. “Think about it; we’ve generated a 10 by 9 correlation
matrix, executing 45 correlation calculations on over a billion rows of data, and the
results were displayed in seconds. By using box plots and correlation matrices, SAS
Visual Analytics can help speed up your analytics life cycle because analytical modelers
can now do variable reduction more quickly and effectively.”
Understand – with Confidence – What It Means
Do your business users need help understanding what all this data exploration is telling
them? No problem. “For business users or analysts who are not familiar with terms such
as ‘regression’ or ‘correlation,’ help is available,” said Marchi. “We’ve added a ‘What
Does It Mean’ feature to explain some of the terminology and help users understand
what they’re seeing.” Click on the information icon, and a pop-up dialog box explains
what the analysis has shown: for instance, that there is a correlation of -0.94, which
suggests there’s a strong linear relationship between two variables.
“By using box plots and
correlation matrices, SAS Visual
Analytics can help speed up
your analytics life cycle because
analytical modelers can now do
variable reduction more quickly
and effectively.”
Justin Choy
Global Product Management,
Business Intelligence, SAS
10. 8
SAS Conclusions Paper
Create Your Own Visual Reports and Dashboards
Once you’ve had a chance to visualize and explore your data, you’ll likely see something
that you want to share with someone else. Or you might want to generate a report
based on the data you have been exploring. The SAS Visual Analytics Designer provides
the tools to create and distribute Web-based reports.
Through a graphical interface, report authors can get wizard-driven help for previewing,
filtering or sampling data before creating visualizations and reports. A variety of formats
are available: standard and three-dimensional bar charts with multiple lines, standard
and three-dimensional pie charts, scatter plots, heat maps, bubble charts, and tile
charts – all with the option for annotated reference lines.
The reports are dynamic and interactive, as Choy demonstrated. “I’ve just opened a
report with three sections, looking at gross margin for product lines and categories over
a range of years. I can click on 2011 to zero in on that year. Click a bar in the yearly
gross margin chart to filter product line and product category gross margin – the other
two segments of the screen display. I can select various product categories to filter on
and get different information – or select a geography from the global filter.”
Choy’s sample report had multiple tabs, each with unique sets of visualizations, all
of them clickable to explore up or down a level in the hierarchy. With drag-and-drop
simplicity, report authors can define the interactions among visuals in a report, specify
the type and appearance of the report page, and add labels and annotation.
Figure 3: Business users can create their own reports and dashboards using a highly
visual workspace.
On-demand ‘What Does It
Mean’ balloons give business
users clear explanations of
complex analytic functions,
correlations and linear
regressions in a language that
they can understand.
11. 9
How to Effectively Realize Data Visualization
Publish Information to the Web and Mobile Devices
Decision makers get the insights they need, when they need them, even when they’re
on the go. Reports are easily read via the Web or on an iPad using the SAS Mobile BI
app, a free iOS app available from the iTunes®
App StoreSM
. Native iPad support uses
the most popular gestures and capabilities, such as zoom, swipe, drag and drop, and
resizing objects on screen. Other mobile devices will be supported in the future.
“Executives or any mobile user can easily benefit from the interactive report viewing
capabilities,” said Marchi, demonstrating on her iPad the reports created earlier in
the demo. “It is highly visual and very self-service for the user – and very simple. Just
open the SAS Mobile BI app to see your portfolio of reports. Click on the library and
navigate to a shared folder to look at reports and download new ones. Click ‘Subscribe’
to download a report. The display shows thumbnails of the reports you subscribe to,
which can be viewed at any time, even when offline.”
The reports offer the same interactions between charts as Choy demonstrated in the
Designer toolbox. Reports can also be saved to PDF or PNG formats for printing.
Figure 4: Get a quick, easy visual understanding of customers or operations on your
mobile device.
“SAS’ strongest play with
SAS Visual Analytics is to
supercharge its predictive
analytics capabilities. The time
shift from hours to minutes
promises to enable customers
to do more analysis with more
data and greater accuracy than
ever before.”
Doug Henschen
InformationWeek (SAS Introduces Big Data
Visual Analytics Platform, March 23, 2012)
12. 10
SAS Conclusions Paper
Closing Thoughts
Even the most common descriptive statistics calculations can become complicated
when you are dealing with big data. You don’t want to be restricted by column limits,
storage constraints and the limited data type support of traditional data architectures.
The answer is an in-memory engine that accelerates the tasks of data exploration and a
graphical interface that displays the results in simple visualizations.
“Dealing with billions of records on a regular basis is just a reality of doing business
today, and SAS understands that,” said Marchi. “SAS Visual Analytics allows you
to explore all of your data using visual techniques combined with industry-leading
analytics. Visualizations such as box plots and correlation matrices help you quickly
understand the composition and relationships in your data.”
Large numbers of users, including those with limited analytical and technical skills, can
quickly view and interact with reports via the Web or mobile devices, while IT maintains
control of the underlying data and security.
The net effect is the ability to accelerate the analytics life cycle and to perform the
process more often, with more data – all the data, if that’s what best serves the
purpose. By using all data that is available, users can look at more options, make more
precise decisions and succeed even faster than before.
“With SAS Visual Analytics, you
can explore your data, discover
new insights, generate reports
and deliver the information to
the people who need it, all from
a single solution. Whatever
your industry – retail, insurance,
health care and life sciences, or
financial services – SAS Visual
Analytics can help.”
Justin Choy
Global Product Management,
Business Intelligence, SAS
Stay Informed
Follow us on twitter:@sasanalytics
Like us on Facebook:SASAnalytics
To get fresh perspectives on
customer analytics from marketing
practitioners writing on the SAS®
CustomerAnalytics blog:
blogs.sas.com/content/
customeranalytics
