Access, Modify, Enhance: Self-Service Data Management in SAS® Visual Analytics

Access, Modify, Enhance: Self-Service Data Management in SAS® Visual Analytics
Paper SAS1877 – 2015
Access, Modify, Enhance: Self-Service Data Management
in SAS® Visual Analytics
Gregor Herrmann and Anand Chitale, SAS Institute Inc., Cary, NC
ABSTRACT
SAS® Visual Analytics provides self-service capabilities for users to analyze, explore, and report on their
own data. As users explore their data, there is always a need to bring in more data sources, create new
variables, combine data from multiple sources, and even periodically update data. SAS Visual Analytics
provides targeted user capabilities to access, modify, and enhance data that is suitable for specific
business needs. This paper provides a clear explanation of these capabilities and suggests best practices
for self-service data management in SAS Visual Analytics including essential performance
considerations.
INTRODUCTION
Sources of data have now gone beyond the boundaries of IT managed enterprise data warehouses, and
organizations are facing the flux of ad hoc sources that business users need to make more informed
decisions.
One of the key criteria for successful use of a BI application for business is having the ability to bring
users’ ad hoc data sources in a self-service manner for data analysis without depending on the IT
resources. Apart from just access to ad hoc data source, there is also an increasing need to enhance
data suitable for the needs of analysis and without the need to request that the IT department make
changes to the centralized data source, which can often take a long time.
Business solutions that allow self-service data access and data manipulation for business analysts are
gaining more and more traction. SAS Visual Analytics comes with capabilities that can empower
specifically enabled end users to bring their own data into the environment and to further refine it by
modifying existing and/or adding new data items. The goal of this self-service data management
capability is to provide a managed, yet self-service way, for users to provision and prepare their own data
without always having to rely on IT. The subsequent sections of this document dive into more details
about these self-service activities.
Please note that SAS Visual Analytics provides data management capabilities for IT as well. However,
those capabilities are not the primary focus of this paper. Please refer to the recommended reading for
more information.
SAS VISUAL ANALYTICS TECHNOLOGY OVERVIEW
SAS Visual Analytics is a BI solution that uses intelligent ways to help business analysts and nontechnical
users visualize data, see patterns and trends, and identify opportunities for further analysis. SAS Visual
Analytics is backed by the power of SAS® Analytics, which is made available to the users in a self-service
1
and approachable manner. SAS Visual Analytics enables the creation and dissemination of dashboards,
reports, and the results of investigative exploration either to the web or to native mobile applications.
Figure 1. Business Visualization Driven by Analytics
SAS Visual Analytics delivers a number of capabilities that run on an enterprise class SAS® 9.4 platform
and that deliver an end-to-end solution for all user needs. SAS Visual Analytics includes the capability to
prepare data prior to making it available to users, an interface for exploring your data (often known as
data discovery), and an interface for building highly interactive and visual reports and dashboards.
Figure 2. SAS Visual Analytics Capabilities Delivered in a Single Solution for Faster and Smarter Decisions
SAS Visual Analytics also includes native applications that run on Android devices or the iPad for the
delivery of information to augment the web. All of this is enabled by the revolutionary and unique SAS
Analytics in-memory engine, SAS® LASR™ Analytic Server, which sits under the surface and can scale
from handling a few hundred megabytes of data to many terabytes. In other words, this is a solution that
can start small and grow with you.
In the sections below, we discuss how SAS Visual Analytics provides self-service approaches for
accessing, modifying, and enhancing data.
2
SELF-SERVICE DATA LOADING
KEY DRIVERS FOR SELF-SERVICE ACCESS TO DATA
As users explore their data and create their own reports, there is always a need to bring in more data
sources, combine data from multiple sources, update data periodically, and in many cases generate new
data variables that are not available in the data source. Over the years, these self-service needs have
become a hot topic and a default expectation by business users using any business intelligence
application. There are two cases where this is particularly important:

Exploring local data files: It is no surprise that business users tend to store valuable information
on their personal PC and would like to incorporate this information wherever it makes sense. This
could be specific lookup information for reference or specific numbers such as the latest currency
conversion numbers or ratio applications that are required to enhance the information appropriate
for business needs.

Adding data sources to the existing data warehouse: There is always a need to bring in additional
information to an existing report. An example could be data that can be accessed via the Internet,
a statistic that has just been released, or social media data, and so on—essentially, anything that
needs to be part of analysis or simply a spreadsheet that you just received from a colleague. It
might also be the case that tables stored in a database are not permanently available for
exploration or report creation.
In both cases, it is of vital importance that the integration of these sources is as easy as possible and is
seamlessly embedded into the workflow of the analyst. SAS Visual Analytics supports this requirement by
enabling you to add data sources at any time during data exploration or report creation within the same
interface.
SUPPORTED FILE FORMATS
When it comes to self-service access to data, it is expected that not all the data will always be coming
from the same source and in the same file format. SAS Visual Analytics supports all file formats of a
regular SAS session. Access to data in SAS is governed by licensed SAS/ACCESS® software engines
that are configured for the system. After access is initiated, only the available SAS/ACCESS engines will
be presented to a user as a choice for available file formats.
Apart from the structured data that is accessible from database systems and local data file formats, SAS
Visual Analytics also supports analyzing unstructured data. You can use the data streams coming from
Twitter with special built-in access to Twitter.com. The only prerequisite is to have a valid Twitter ID.
Figure 3 shows all the accessible data sources and their supported file formats.
3
Figure 3. Import Data Dialog Showing a Choice of Supported File Formats
* The upcoming release of SAS Visual Analytics will add Facebook and Google Analytics and more
Hadoop versions to the list of available data sources.
AD HOC DATA ACCESS IN SAS® VISUAL ANALYTICS DESIGNER OR SAS® VISUAL ANALYTICS
EXPLORER
In both SAS Visual Analytics Designer and SAS Visual Analytics Explorer, you can add data whenever
necessary via the Add Data Source window:
4
Figure 4. Add Data Source Window in SAS Visual Analytics Designer with Accessible Data Sources
The range of data sources that analysts want to explore or include in their reports has already broadened
in recent years, and it can be expected that it will further grow in the future. SAS Visual Analytics supports
this need by enabling the user to access multi-structured data for advanced analytics.
The user interface groups the distinctive data sources as Local, Server, and Social Media.

Local is everything available on a user’s PC or laptop, where you use SAS Visual Analytics.

Server is the machine where data sources reside. Depending on your file format, it might be
required to enter credential information to connect to the respective data source. For example, if
you want to load a database table coming from an RDBMS system like Oracle, the user interface
will present the option to provide database access user credentials and data table to access.
Server also lists the available Hadoop sources that can be accessed via SAS/ACCESS ® Interface
to Hadoop, which comes with SAS Visual Analytics, if you are using Hadoop as your data
persistence layer. SAS/ACCESS Interface to Hadoop lets you interact with your data using SQL
constructs through Hive and HiveServer2. It also lets you access data directly from the underlying
data storage layer, the Hadoop Distributed File System (HDFS). This differs from the traditional
SAS/ACCESS engine behavior, which exclusively uses database SQL to read and write data.

Social Media provides the option to create a connection to Twitter.com with a valid Twitter ID.
If your data is stored locally, it must be uploaded to the SAS LASR Analytic Server (in memory) to be
further accessible inside SAS Visual Analytics. Because uploading these data files might take some time
depending on file size and the bandwidth of network connection, an administrator can limit the size of files
that can be uploaded to avoid user frustration. Nevertheless, files of a reasonable size (up to couple of
megabytes) will be uploaded quickly.
SELF-SERVICE DATA ACCESS VIA THE AUTO LOAD MECHANISM
Another option of self-service data loading is implemented in SAS Visual Analytics via the auto load
functionality. The principle is quite simple: An administrator defines a location where the users can drop
5
the data files that they need to be loaded in memory for analysis. A system service checks periodically if
data has been dropped in that area, and, if available, the data file gets uploaded automatically into
memory.
To enable this degree of automation in the data loading process, there are some rules to follow:
1. The auto load mechanism supports only SAS data sets, Excel sheets, CSV files, and delimited
text files.
2. When Excel sheets or CSV files are being loaded, variable names have to be in the first row of
the document.
The auto-loading mechanism also supports appending new data to existing data tables in memory. There
is a special Append folder that handles delta loading, and is configured during the auto load configuration.
You can even unload unwanted data by placing it in the Unload folder. Auto load is always tied to a
specific library.
Figure 5. Default Structure of the AutoLoad Directory
By default, the Visual Analytics Public LASR library is enabled for auto loading. Individual behavior of the
auto load mechanism can be controlled via a couple of extended attributes:
6
Figure 6. Extended Attributes of the Visual Analytics Public LASR Library
Auto load mechanism is useful when you want to load larger data files to SAS LASR Analytic Server and
want to avoid the browser limitations of the size of data files when importing data using the user interface.
If security requirements do not allow one generic auto load location for every department you can set up
several auto load instances. Thus, you can also set individual thresholds for the memory consumptions of
those auto load instances by assigning a separate LASR Analytic Server for each.
*In an upcoming release of SAS Visual Analytics, there will be two important enhancements to auto-load:

An administrator can enable data compression when loading data into memory

Text files with a delimiter will be supported
ENHANCING DATA
CALCULATE NEW DATA ITEMS
As users start using the centrally provided data sources, or even bring their own additional data source,
there are business needs to build logical calculations on the fly using these data sources. In traditional BI
systems with a complete IT-governed strategy, a user would have to request IT to build the new required
calculations into the data source, which would take its own time based on priorities and data management
schedules.
With self-service approaches supported by new data visualization solutions, users can leverage the builtin capability of defining new variables on the fly, and build their own business logic to test the outcomes in
some cases as a playpen setup.
SAS Visual Analytics as a self-service application provides several capabilities to create new data items
on the fly. These can be classified into four main categories:
1. Calculated items on detail data. Here the calculation will be done before any aggregation
happens.
7
Example: X = Sum (A/B)
2. Calculate aggregated measures. In this case, data gets aggregated first before the calculation
is done.
Example: X = Sum (A) / Sum (B)
3. Parameterized calculations either on detail or aggregated level. Here, the resulting value of a
calculation depends on the value of the parameter passed to the calculation expression.
Example: Parameter Y = 1,2,3,4….
X=A+B*Y
4. Creating custom categories. This enables users to combine the values of a variable and create
new variables for data presentaion.
Example: Creating a new region as “Nordics,” which combines the values for regions = Norway,
Denmark, Sweden, and Finland all together
Here are some important things to notice:

The result of a calculated item on the detail level can be of any data type (character, numeric,
date, time, or datetime).

The result of an aggregated calculation is always numeric.

Parameterized calculations can be done both on the detail or aggregated level.
The combination of the capabilities described above makes it easy for someone designing a report or
analyzing data in SAS Visual Analytics to create new data items as required. Particularly of importance is
the use of parameterized calculations, which not only provides an advanced way to interact with data, but
also helps build scenario-driven analysis for the report users.
As shown in Figure 7 below, you can choose from a wide variety of operators ranging from simple
arithmetic computations to advanced text manipulations. This capability makes SAS Visual Analytics
attractive for users with different skill levels to create new calculations of varying complex business logics
to be created using simple interactive drag-and-drop actions and without the need of special
programming skills.
8
Figure 7. New Aggregated Measure Using Parameter for Dynamic Computation
Figure 8 shows how the parameter-driven calculation “Projected Order” (as calculated in Figure 7) is used
in a report, enabling the user to run a what-if analysis using the varying values of “Advertising Effect.”
Figure 8. Using a New Calculated Measure in a What-If Style Scenario Controlled via a Slider
USING PARAMETERS TO MAKE REPORTS DYNAMIC
In SAS Visual Analytics 7.1, parameters were introduced. You can use parameters in calculations, display
rules, filters, and ranks in SAS Visual Analytics Designer. Parameters enable a report designer to fulfill
complex report design requirements that can give the report consumer a more dynamic reporting
experience. Before going into detail on how parameters can be used in a report, let us first look at a few
basic facts about parameters:
9

Parameter types can be numeric or character. Date type parameters are not yet supported.

Numeric parameters require a current value, a maximum value, and a minimum value during creation.
Specifying a format is optional.

Character parameters do not require a value when created.

Parameters can be edited and deleted from the data pane.

You cannot delete a parameter that is being used in any object.
Figure 9. Edit Parameter Window Showing a Numeric Parameter
Listed below are some possible scenarios where the use of parameters can be beneficial:
1. Parameters can be used to dynamically influence the result of a calculation as already shown in
Figure 8. This is of great value for an information consumer as he or she can play around with these
values and instantly see the effect in a graph or tabular object.
2. Display rules usually have fixed thresholds. When you use parameters within the definition of a
display rule, you can now modify the coloring of values dynamically as shown in Figure 10. Another
possible scenario is when you display a sum in a crosstab that contains a hierarchy. Different
hierarchy levels require different thresholds for coloring the values and can be easily achieved using
rules based on parameter-driven variables.
3. Adding flexibility for advanced filtering: Filtering high cardinality categories often requires wildcards,
because the report consumer might not know the exact value of a data item or declines to enter the
complete value when entering only a part of it can deliver the same filter result. Use of parameters
allow for creation on more complex filter conditions.
4. Cascading prompts for section-level or report-level filters: Cascading prompts can be generated in
SAS Visual Analytics Designer out-of-the-box, if the filter objects are part of the report body and when
using interactive links between filter controls. With parameter-driven values in filter controls, a similar
cascading behavior can be achieved when using filter controls for section- and report-level prompts.
5. Ranking: With the help of parameters, we can allow users to enter the value for Top N/Bottom N
dynamically. If you have a visual showing Top 3 (as a graph or list table), it might be interesting from
a business perspective to see who the next best competitor is.
10
Figure 10. Using a Parameter to Modify Display Rules via Slider
These are only a few suggestions on how parameters can be used to meet complex report design
requirements. I am sure there are a lot more. If you are encountering a problem during design of your
report, check if using parameters might solve your problem.
DATA PREPARATION IN SAS® VISUAL DATA BUILDER
So far, we have only considered loading data into SAS Visual Analytics and enhancing it from within the
interface. But there might be cases where an advanced business user with appropriate skills would want
to join several data sources before loading into memory. The interface that supports these types of
activities in SAS Visual Analytics is called SAS Visual Data Builder. It has been designed to fulfill basic
data preparation requirements to bring the data into a shape that is effective in analytical data discovery
and reporting. Behind the scenes, SAS Visual Data Builder generates SAS® code that can be executed in
a separate SAS session.
In SAS Visual Data Builder, you have three main choices:
1. Upload data as is: This is identical to what has been described in the self-service data loading
section.
2. Create a data query: This allows you to join several files and compute additional columns if
necessary. The resulting data set can be stored permanently to a database or loaded to SAS
LASR Analytic Server in memory directly for instant access. As part of query builder, users can
create new columns by using the advanced expression builder user interface that comes with a
library of pre-defined functions and allows users to add or update business labels and formats for
their database columns,that are useful down the line for users. (See Figure 9.)
3. Create a Star Schema: Those working with traditional BI systems would be familiar with Star
Schema—which is essentially a view of data coming from a central fact table that is associated
with multiple dimension tables. Using a star-schema approach can be beneficial in cases where
you want to upload only a specific view of your entire data into memory. This approach can help
save some memory space in comparison to loading the complete data tables into memory.
11
Figure 11. Creating a Query in SAS Visual Data Builder
When modeling your data in the form of a Star Schema, certain aspects need to be followed:
1. Only one fact table is allowed.
2. One variable in the dimension table defines the relation to one variable in the fact table.
3. Dimension keys must be unique.
Figure 12. Star Schema with One Fact Table and Three Dimension Tables
The default output format is a view. (See the checkbox ticked in Figure 12.) When you create a Star
Schema with output format view, the fact and dimension tables will be loaded to the LASR Analytic
Server. However, the result table will not be materialized immediately. The logical join will only be
12
executed when the view is being used by a report or exploration, thus saving valuable memory space for
other tables or actions.
PERFORMANCE CONSIDERATIONS
Data preparation and its management are critical activities for a successful running of any enterpriseclass business intelligence system. To make the right choice and provide an optimal performing
environment, some key factors must be taken into consideration. Some of these are more important from
the system architecture perspective as well. Here are a few cases for your consideration:
1. All tables being used are SAS data sets. In this case, be aware that data preparation steps will be
SAS programs running on the workspace server of the SAS Visual Analytics environment. This limits
both computing and storage resources that are available to the environment and thus calls for
appropriate sizing of the SAS Visual Analytics environment. Be aware that when you run joins on that
environment, additional temporary space will be needed to complete these steps successfully. It is
also required that before running the code, all necessary SAS data sets should be copied to the
environment to avoid data movement across networks. This adversely affects performance.
2. All tables are stored in one relational database. In this case, the generated SQL code will be
passed to the underlying database for processing (implicit pass-through) when the resulting table will
also be stored in the database itself. An additional step is then required to load the resulting data
table from the database to SAS LASR Analytic Server. If the database supports parallel load with the
SAS Embedded Process, loading of the result table to SAS LASR Analytic Server (in memory) will be
reasonably faster.
3. A mixture of SAS data sets and tables in a relational database. This is the most complex case. If
possible, joins should be run on the database tables. This leverages the power of the database itself.
The resulting data table can then be copied to the SAS workspace server to run remaining data
preparation steps using SAS.
4. All tables are stored in Hadoop. In this case, the SAS/ACCESS Interface to Hadoop can be used to
pass the processing to the Hadoop distributed environment. If you are running an asymmetric
architecture (SAS LASR Analytic Server and Hadoop running on separate nodes) and the Hadoop
version used supports the SAS Embedded Process, parallel loading can be used to make tables
available in LASR. In this case, it is recommended to run data preparation completely in Hadoop
especially if data volumes are big.
CONCLUSION
With the current shift from the use of centralized enterprise data warehouses, exclusively owned and
operated by IT and supporting traditional static reporting to data discovery tools that provide self-service
data management capabilities to the business, it is obvious that data preparation capabilities will be
introduced on a broader scale across the userbase in maturing organizations. SAS Visual Analytics today
already supports this shift and will extend these capabilities in future releases.
RECOMMENDED READING
Atassi, Sam, and Malcom Alexander. 2013. “Self-Service Data Management: SAS Visual Data Builder.”
Proceedings of the SAS Global Forum 2013 Conference. Cary, NC. SAS Institute Inc. Available at
http://support.sas.com/resources/papers/proceedings13/051-2013.pdf.
Mehler, Gary, and Donna Bennett. 2014. “Big Data Everywhere! Easily Loading and Managing Your Data
in the SAS® LASR™ Analytic Server.” Proceedings of the SAS Global Forum 2014 Conference. Cary,
NC. SAS Institute Inc. Available at http://support.sas.com/resources/papers/proceedings14/SAS3472014.pdf.
Rivers, Kerri, and Christopher Redpath. 2015. “Tips and Techniques for Efficiently Updating and Loading
Data into SAS® Visual Analytics.” Proceedings of the SAS Global Forum 2015 Conference. Cary, NC.
SAS Institute Inc.
13
SAS Institute Inc. 2014. SAS Visual Analytics: User’s Guide. Cary, NC: SAS Institute Inc.
CONTACT INFORMATION
Your comments and questions are valued and encouraged. Contact the author at:
Gregor Herrmann
SAS Institute Inc.
SAS Campus Drive
Cary, NC, 27513
[email protected]
Anand Chitale
SAS Institute Inc.
SAS Campus Drive
Cary, NC, 27513
[email protected]
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of
SAS Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies.
14
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement