Data Mining From A to Z:
How to Discover Insights and Drive Better Opportunities
Analytical Life Cycle: Combining Data,
Discovery and Deployment ............................................2
What Can Data Mining
Help You Discover? ..........................................................3
A Closer Look at the Role of Data Mining
in the Discovery Process .................................................5
Step 1: Turn a Business Question Into
an Analytical Hypothesis ......................................................5
Step 2: Prepare the Data for Data Mining .......................5
Step 3: Explore the Data ......................................................5
Step 4: Model the Data ........................................................6
Data Mining Solutions ...........................................6
Data Mining and Machine Learning ..................................6
Factory Miner for an
Automated Approach to Data Mining ..............................8
Scaling Your Discovery Process to Handle
Big Data and Complex Problems ......................................8
Integration Eases Model Deployment,
Monitoring and Management ...........................................10
Learn More .......................................................................11
So much data and multitudes of decisions. Organizations
everywhere struggle with this dilemma. The data is growing,
but what about your ability to make decisions based on those
huge volumes of data? Is that growing too? For many, unfortu-
nately, the answer is no.
Data pours in at unprecedented speeds and volumes from
everywhere. But making fact-based decisions is not dependent
on the amount of data you have. Actually, having so much data
can be a paralyzing factor. Where do you even begin? Your
success will depend on how quickly you can discover insights
from all that data and use those insights to drive better actions
across your entire organization.
That’s where predictive analytics, data mining, machine learning
and decision management come into play. Predictive analytics
helps assess what will happen in the future. Data mining looks
for hidden patterns in data that can be used to predict future
behavior. Businesses, scientists and governments have used this
approach for years to transform data into proactive insights.
Decision management turns those insights into actions that are
used in your operational processes. So while the same
approaches can still be applied today – they need to happen
faster and at a larger scale, using the most modern techniques
Forward-thinking organizations use data mining and predictive
analytics to detect fraud and cybersecurity issues, manage risk,
anticipate resource demands, increase response rates for
marketing campaigns, generate next-best offers, curb customer
attrition and identify adverse drug effects during clinical trials,
among many other things.
Because they can produce predictive insights from large and
diverse data, the technologies of data mining, machine learning
and advanced analytical modeling are essential for identifying
the factors that can improve organizational performance and,
when automated in everyday decisions, create competitive
advantage. And with more of everything these days (data,
computing power, business questions, risks and consumers),
the ability to scale your analytical power is essential for staying
ahead of your competition.
Deploying analytical insights quickly ensures that the timeliness
of your analytical models is not lost due to slow processes like
rewriting code for each environment, revalidating the rewritten
models and other manual processes. If you can rapidly deploy
your analytical models, the context and relevance of the models
is not lost and you retain competitive advantage.
So how do you create an environment that can help your orga-
nization deal with all of the data being collected, all of the
models being created and all of the decisions that need to be
made, all at an increasing scale? The answer is an iterative
analytical life cycle that brings together:
• Data – the foundation for decisions.
• Discovery – the process of identifying new insights in data.
• Deployment – the process of using newly found insights to
drive improved actions.
Figure 1: An integrated combination of data, discovery and
deployment is needed to derive and put into action the fast
insights needed for scalable decision making.
Analytical Life Cycle:
Combining Data, Discovery and
Even though the majority of this paper is focused on using data
mining for insights discovery, let’s take a quick look at the entire
iterative analytical life cycle, because that’s what makes predic-
tive discovery achievable and the actions from it more valuable.
• Ask a business question. It all starts here. The discovery
process is driven by asking business questions that produce
innovation. This step is focused on exploring what you need
to know, and how you can apply predictive analytics to your
data to solve a problem or improve a process.
• Prepare data. Collecting data certainly isn’t a problem these
days – it’s streaming in from everywhere. Technologies like
Hadoop and faster, cheaper computers have made it
possible to store and use more data, and more types of data,
than ever before. But there is still the issue of joining data in
different forms from different sources and the need to trans-
form raw data into data that can used as input for data
mining. Data scientists still spend much of their time dealing
with these tasks.
• Explore the data. Interactive, self-service visualization tools
need to serve a wide range of user personas in an organiza-
tion (from the business analyst with no analytical knowledge
to a data scientist) to allow searches for relationships, trends
and patterns to gain deeper understanding of the informa-
tion captured by variables in the data. In this step, the
hypothesis formed in the initial phase of the project will be
refined and ideas on how to address the business problem
from an analytical perspective are developed and tested.
While examining your data, you may find the need to create,
select or transform some data to create more precisely
focused models. Fast, interactive tools help make this an iter-
ative process, which is crucial for identifying the best ques-
tions and answers.
• Model the data. In this stage, the data scientist applies
numerous analytical modeling algorithms to the data to find
a robust representation of the relationships in the data that
help answer the business question. Analytical tools search for
a combination of data and modeling techniques that reliably
predict a desired outcome. Experimentation is key to finding
the most reliable answer, and automated model building can
help minimize the time to results and boost the productivity
of analytical teams. In the past, with manual model-building
tools, data miners and data scientists were able to create
several models in a week or month. Today, they can create
hundreds or even thousands. But how can they quickly and
reliably find the one model (out of many) that performs best?
With automated tournaments of machine-learning algo-
rithms and a clearly defined champion model, this has
become an easy process. Analysts and data scientists can
now spend their time focusing on more strategic questions
• Implement the models. Here we move from the discovery
phase to deployment – taking the insights learned and
putting them into action using repeatable, automated
processes. In many organizations this is the point where the
process often slows down dramatically because there is no
defined handshake between the two worlds of discovery
and deployment, let alone automation. Bringing these two
worlds together to create an integrated transition helps
decrease time to value for predictive analytics. The faster
your business can use the answers generated by predictive
analytics for better decision making, the more value will be
generated. And, a transparent process is important for
everyone – especially auditors.
• Act on the new information. There are two types of decisions
that can be made based on analytical results. Strategic deci-
sions are made by humans who look at results and take action.
Operational decisions are automated – like credit scores or
recommended best offers – and don’t require human inter-
vention. More and more organizations are looking to
automate operational decisions and provide real-time
answers and results to reduce decision latencies. Basing oper-
ational decisions on answers from analytical models also
makes decisions objective, repeatable and measurable. The
integration with enterprise decision management tools
enables organizations to build comprehensive and complete
operational decision flows that combine data-driven analytics
and business rules for optimal automated decisions.
• Evaluate your results. The next – and perhaps most impor-
tant – step is to evaluate the outcome of the actions
produced by the analytical model. Did your predictive
models produce tangible results, such as increased revenue
or decreased costs? With continuous monitoring and
measurement of the models’ performance, you can evaluate
the success of these assets and make sure they continue to
produce the desired results.
• Ask again. Because your data is always growing and ever
changing, relationships in data that your models use for
predictions also change over time. Constant evaluation of
your analytical results will identify the degradation of model
accuracy. Even the most accurate models will have to be
refreshed over time, and organizations will need to go
through the discovery and deployment steps again. It’s a
constant and evolving process.
SAS provides an integrated, complete analytics platform that
handles every step in the iterative analytical life cycle. This
remainder of this paper will focus on the data discovery portion
of the life cycle – and the data mining tools you’ll need to
quickly build the most accurate predictive models possible.
What Can Data Mining
Help You Discover?
Data mining provides a core set of technologies that help orga-
nizations anticipate future outcomes, discover new opportuni-
ties and improve business performance. It can be applied to a
variety of customer issues in any industry – from customer
segmentation and targeting, to fraud detection and credit risk
scoring, to identifying adverse drug effects during clinical trials.
A common use of data mining and machine-learning tech-
niques is to automatically segment customers by behavior,
demographics or attitudes – to better understand needs of
specific groups and serve them in a more targeted way. This
analytical segmentation, or unsupervised modeling, helps to
identify groups of customers that are similar and might react to
Figure 2: The analytical life cycle is an iterative process of making discoveries from your data and applying new insights to
continually improve predictive models and their results.
certain offers or activities in a similar way. Using these segments,
you can create models for each group to predict the next-best
offer or activity to which they’re most likely to respond. To ensure
that you only engage desired customers, you can further comple-
ment the customer acquisition model with a risk-scoring model
to find out who is a good credit risk and actually worth the invest-
ment to acquire or retain.
Another important use for data mining and machine learning is
to help detect fraud, which is important as fraudsters become
more sophisticated in their tactics. Models can be built to cross-
reference data from a variety of sources, correlating nonobvious
variables with known traits to identify new patterns of fraudulent
Because of its potential to produce accurate predictive insights
from huge volumes of diverse data, data mining has proven to
be an invaluable component of many analytical initiatives. Data
mining and machine learning can help you:
• Automatically discover patterns, trends and relationships
represented in data.
• Develop models to better understand and describe charac-
teristics and activities based on these patterns.
• Use those insights to help evaluate future options and make
• Create score code that expresses the calculations to be
made for timely, appropriate actions.
Common Applications for Data Mining Across Industries
What Is Predicted?
How to better target product/service offers?
Profiling and segmentation.
Customer behaviors and needs by segment.
Which product/service to recommend?
Cross-sell and up-sell.
Probable customer purchases.
How to grow and maintain valuable customers?
Acquisition and retention.
Customer preferences and purchase patterns.
How to direct the right offer to the right person
at the right time?
The success of customer communications.
Which customers to invest in and how to best
appeal to them?
Profitability and lifetime value.
Drivers of future value (margin and retention).
Industry-Specific Data Mining Applications
What Is Predicted?
How to assess and control risk within existing
(or new) consumer portfolios?
Credit scoring (banking).
Creditworthiness of new and existing sets of
How to increase sales with cross-sell/up-sell,
loyalty programs and promotions?
Recommendation systems (online retail).
Products that are likely to be purchased next.
How to minimize operational disruptions and
Asset maintenance (utilities, manufacturing, oil
The real drivers of asset or equipment failure.
How to reduce health care costs and satisfy
Health and condition management (health
Patients at risk of chronic, treatable/prevent-
How to decrease fraud losses and lower false
Fraud management and cybersecurity (gov-
ernment, insurance, banks).
Unknown fraud cases and future risks.
How to bring drugs to the marketplace quickly
Drug discovery (life sciences).
Compounds that have desirable effects.
In some cases to expedite modeling processes, you may want
to sample the data – that is, create a smaller subset of the data
that represents the target data set. Data mining can only
uncover patterns already present in the data, so the sample
should be representative and large enough to contain the
significant information. The analytics base table is also generally
divided into at least two sets: the training set and the test set.
The training set is used to train the data mining and machine-
learning algorithm(s), while the test set is used to verify the
accuracy of any patterns found.
Step 3: Explore the Data
Next, you’ll want to explore the data and search for anticipated
relationships, unanticipated trends and anomalies to gain an
understanding of the information you’re working with and
further refine ideas and questions. Data exploration can also
help pinpoint data quality problems such as data errors, missing
values or data distributions that need to be transformed for the
modeling stage. In addition, you can use several other types of
techniques to detect patterns in the data that can help you build
more accurate predictive models or help you create additional
input data for your predictive model.
• Clustering (or unsupervised modeling) identifies groups or
structures in the data that are similar, beyond the structures
otherwise visible in the data.
• Association-rule learning searches for relationships among
variables, such as products frequently bought together
(known as market basket analysis), which can lead to further
recommendations for purchase.
• Text analytics can help you to create new structured informa-
tion from electronic text data. This new data can help to
improve the accuracy of your models. For example, inte-
grating customer comments on your products and services
from call center notes or reviews on social media forums
often produces more accurate churn prediction models.
• Interactive data visualization presents results graphically and
lets users interact with these graphs to more easily identify
important patterns or anomalies with the data that might
have an impact in the model-building stage.
Often, you’ll need to modify your data before modeling so you
should plan on a step for creating, selecting and transforming
variables to focus your model-selection process. Based on your
discoveries in the exploration phase, you may need to manipu-
late your data to introduce new variables, fill in missing values or
look for outliers so you can reduce the number of variables to
only the most significant ones.
A Closer Look at the Role of Data
Mining in the Discovery Process
Data mining and machine learning lie at the heart of the
discovery process. But there’s more to discovery than just
building an analytical model. You’ll get better results if you take
an iterative, holistic approach.
Step 1: Turn a Business Question Into an
The first step in the discovery process is to ask a business
question (see tables on page 4). Usually an organization has a
general idea of what it wants to achieve – something like, “We
want to reduce the churn of our valuable customers.” To
address these issues with analytics, business questions must be
specified in detail or transformed into an analytical hypothesis.
For example, every predictive model requires a well-defined
outcome, a label or target. If you want to predict customer
churn, you need to define churn as an outcome for the model.
However, churn is likely defined differently in different organiza-
tions. Does it refer to someone actively canceling a contract or
someone who is dormant in his activities? How long can a
customer remain dormant before being classified as a churner?
What is valuable? Do we include only historical value or poten-
tial future value (lifetime value) of a customer? Your first step in
the discovery process is to identify an issue and translate the
issue into a question that can be addressed with analytics.
Step 2: Prepare the Data for Data Mining
To begin, you must determine what data is needed to answer
the question. Based on the specifics of the business question,
an analyst evaluates the data that is available and decides if the
data has the potential to answer the question at hand. If not,
external data may be needed or new data might need to be
collected. Often, the data is in different systems and needs to be
accessed and turned into a data set that can be used for data
mining and machine learning. Predictive or supervised models
require a single record per entity to model. (An analytics base
table for forecasting or market analysis will look different from a
table for predictive or supervised modeling). If you want to
model the likelihood of customer churn, you need to create a
single table where each record contains all the data attributes
for a single customer. This often requires a significant amount of
data aggregation and transformation. Once a single analytics
base table for the analysis has been aggregated, the other
aspects of the life cycle come into play. Because it is necessary
to experiment with data, the preparation stage is also very itera-
tive with the analyst trying different types of data to get the most
accurate predictive results.
are big timesavers here. When you are satisfied with the results
of your modeling endeavors, you then begin the deployment
process. But because it’s a completely iterative process, there
are constant examinations and adjustments. As discussed
before, there are several steps involved in the deployment
process (see the SAS Analytical Life Cycle section on page 2).
For more information on the deployment process, read
Data to Decision: How SAS® Decision Manager Automates
. To learn more about data mining and
discovery, keep reading!
Data Mining Solutions
Data mining and machine learning enable you to discover
insights that drive better decision making. With SAS data mining
solutions, you can streamline the discovery process to develop
models quickly so you can understand key relationships and
find the patterns that matter the most.
Data Mining and Machine Learning
SAS Enterprise Miner is a comprehensive, graphical workbench
for data mining. This widely acclaimed and extensive platform
provides capabilities to prepare data for predictive analytics,
identify the most significant variables, develop models using
the most modern data mining and machine-learning algo-
rithms, easily validate the accuracy and fitness of the model(s),
and generate assets that allow a simple deployment of analyt-
ical models into your operational applications for automated
Powerful data preparation tools address data quality problems,
such as missing values and outliers, and help you develop
segmentation rules. Interactive data exploration enables users
to create dynamic, linked plots to identify relationships within
the data. SAS Enterprise Miner provides dozens of advanced
statistical and machine-learning algorithms for descriptive and
predictive modeling, including clustering, link and market
basket analysis, principal component analysis, decision trees,
bagging and boosting, Bayesian networks, neural networks,
random forests, linear regression, logistic regression, support
vector machine, time series data mining and many more.
At the end of the model development pipeline, complete, opti-
mized scoring code is delivered for easy deployment of the
unsupervised or supervised models in SAS, C, Java and PMML
for scoring data in SAS as well as in other environments. Score
code can also be delivered automatically as an in-database
Step 4: Model the Data
After carefully exploring and preparing your input data, you are
ready to create predictive or supervised models to search for a
combination of the data that reliably predicts a desired
outcome. Depending on the data and issue at hand, you can
choose from a variety of modern machine-learning and statis-
tical techniques to solve your problem – including classification,
regression, neural networks, random forests, support vector
machine, incremental response or time series data mining – as
well as industry-specific techniques such as credit scoring in
banking or rate making for insurance.
The selection of the most appropriate techniques depends on
several factors. Is it more important to have a model that
predicts your desired outcome with the highest accuracy or is it
also (or even more) important to have transparency into the
data relationships that drive the predictions? Automated
machine-learning techniques are often too complex to allow the
exploration of business drivers from the model results, while
other statistical techniques such as regression or decision trees
are more transparent and are preferred in regulated industries.
To get the most value from your predictive models, you’ll want
to constantly evaluate the usefulness and reliability of the
findings from your data mining processes. Not all patterns
found by the data mining algorithms will be valid. The algo-
rithms might find patterns in the training data set that are not
present in the general data set. (This is called overfitting.) To
address this concern, patterns are validated against a test set of
data. The patterns learned on the training data will be applied to
the test set, and the resulting output is compared to the desired
(or known) output.
For example, a data mining algorithm that had been trained to
distinguish fraudulent credit card transactions from legitimate
ones would then be applied to the test set of transactions on
which it had not been trained. The accuracy of the patterns can
then be measured from how many credit card transactions are
correctly classified. If the learned patterns do not meet desired
standards, modifications are made to the preprocessing and
data mining techniques until the result is satisfactory and the
learned patterns can be successfully applied to operational
Data scientists and data miners need to experiment with a multi-
tude of predictive modeling and machine-learning algorithms
in order to find the one that works best for their specific
problem. Automated modeling tournaments where users can
experiment to identify the winning modeling strategy quickly
function for scoring inside Hadoop as well as industry-leading
databases such as Teradata, IBM, Oracle, Pivotal, Aster Data,
SAP HANA, etc., for very seamless integration with business
applications and fast operational results.
In addition to generating score code in different languages and
formats, SAS Enterprise Miner also generates many assets that
enable easy deployment, management and monitoring of
predictive models as part of operational business processes.
All of these assets are supported by metadata to provide mean-
ingful documentation around the entire process.
The SAS Enterprise Miner data mining process is driven by a
process flow diagram that you can modify, save and share. The
drag-and-drop GUI enables business analysts with little statis-
tical expertise to navigate through the data mining process,
while the quantitative expert can go behind the scenes to fine-
tune the analytical models.
With SAS Enterprise Miner, you can:
• Create training and test sample data sets with
high predictive value.
• Interactively explore relationships and anomalies
in the data.
• Create, transform and select the most appropriate
variables for analysis.
• Apply a range of modeling techniques to identify
patterns in the data.
• Validate the usefulness and reliability of findings
from the data mining process.
• Create all required assets for easy model deploy-
ment, monitoring and management.
Figure 3: Decision trees are just one of the many modeling techniques included with SAS Enterprise Miner. They can be developed
interactively or in batch mode. Numerous assessment plots help gauge overall tree stability.
as the producer of organizational best-practice modeling pipe-
lines for different projects and other users of the environment
consume these best practices in a self-service fashion for
And SAS Factory Miner does not stop with the identification of a
champion model for each segment. Complete code is auto-
matically created for the entire scoring pipeline (including data
transformations) of each model for deployment in SAS or other
environments, such as databases or Hadoop.
In addition, all model development and scoring assets can be
registered to SAS Decision Manager, a centralized web-based
environment for managing the life cycle and governance of
your modeling assets from SAS or third-party providers,
including open-source analytics.
The automation, ease of use, scalability and collaboration capa-
bilities of SAS Factory Miner ramp up your predictive model-
building power, increase the productivity of your analytics staff,
enable collaboration across dispersed analytics teams, as well
as expand your analytics talent pool through the democratiza-
tion of machine-learning techniques.
Scaling Your Discovery Process to Handle
Big Data and Complex Problems
Big data and complex problems call for big analytics solutions.
At SAS, we amp up your discovery power with distributed
in-memory analytics. The idea is simple yet powerful. Break your
data into smaller chunks and distribute the volume of the data
and the complexity of the problem across your compute
Factory Miner for an
Automated Approach to Data Mining
As organizations apply more targeted analytics to their growing
number of customer and business segments, there is a need to
create even more predictive models at more granular levels. For
example, instead of developing one model for the entire
customer base, marketing departments want to create specific
models for many customer segments. A retailer may want to
develop cross-sell models for a large number of product cate-
gories. Or, a transport enterprise will want to build predictive
maintenance models for different components of the vehicles it
has in operation. And while this makes it necessary to create a
lot more models, most analysts and data scientists don’t have
the luxury of more time.
With SAS Factory Miner, you get an interactive predictive
modeling environment that makes it extremely easy to create,
modify and assess hundreds, or even thousands, of models very
quickly. With just a few clicks, you can access, modify and trans-
form your data, choose which machine-learning techniques you
want to apply and run the models in an automated model tour-
nament environment to quickly identify the best performer for
each segment. Modeling techniques included in SAS Factory
• Bayesian networks.
• Decision trees.
• Gradient boosting.
• Neural networks.
• Random forests.
• Support vector machines.
• Generalized linear models.
• Linear regression.
• Logistic regression.
Users can easily identify modeling exceptions (segments where
the automated approach does not generate models that meet
acceptance criteria). The white-box design of SAS Factory Miner
lets users easily modify predictive modeling pipelines and fine-
tune parameters of pipeline components for better results
where required. They can even create their own customized
modeling pipelines for their favorite analytical projects,
including data preparation, feature engineering and selection
and learning algorithms, and share them with other users to
create a repository of organizational best practices. This collabo-
ration across the entire organization can help expand the
analytics talent pool in your organization. The data scientist acts
With SAS Factory Miner, you can:
• Boost discovery productivity.
• Automate model development.
• Explore new ideas faster.
• Collaborate with your analytics peers across
• Expand your analytics talent pool through
automated self-service machine learning.
• Put large predictive model portfolios into produc-
tion more efficiently and manage them with ease.
engines, whether it’s on a single machine with a multitude of
processing cores (CPUs) or a network of computers, such as a
Hadoop cluster. The processing is done entirely in memory
whenever possible, including the communication between the
processing units (CPUs), which makes this process really fast.
SAS distributed, in-memory analytics processing takes advan-
tage of a highly scalable and reliable analytics infrastructure –
including database appliances like Pivotal Greenplum, Teradata,
Oracle and SAP HANA – and commodity hardware using open
source Hadoop or Hadoop Cloudera and Hortonworks distribu-
tions. For the users, nothing much changes. They can work from
the same familiar interface for their data mining, predictive
analytics and machine-learning projects, while SAS In-Memory
Analytics takes care of the optimal workload distribution on the
Figure 4: Customizable assessment techniques in SAS Factory Miner enable you to generate champion models for every segment
in your data.
SAS High-Performance Data Mining lets you analyze large
volumes of diverse data using a drag-and-drop interface and
powerful descriptive, predictive and machine-learning methods.
A variety of modeling techniques – including random forests,
support vector machines, neural networks and clustering – are
combined with data preparation, data exploration and scoring
capabilities. Because you’re able to build and run more models
faster, you can ask more questions and bring new ideas into your
data mining process. SAS High-Performance Text Mining lets you
gain quick insights from large unstructured data collections
involving millions of documents, emails, notes, report snippets,
social media sources, etc. Support is included for parsing, entity
extraction, automatic stemming and synonym detection, topic
discovery and singular value decomposition (SVD). Text mining
results can be used as inputs into high-performance data mining
to improve your predictive modeling power.
Furthermore, business rules are being used in conjunction with
analytical models to make decisions more flexible and agile.
With SAS Decision Manager, business rules help define the
actions based on specific conditions in business processes.
In the past, the deployment of a predictive model into the
production environment has been manually performed by IT,
often resulting in huge delays before the model could be used.
With constantly changing market conditions and new data
continuously arriving, it’s possible for models to become
obsolete before they are even deployed. With the seamless
integration of the discovery and deployment phases of the
analytical life cycle, SAS enables organizations to automate this
process. SAS Decision Manager provides a streamlined inter-
face to deploy models to execution environments in real time or
batch without recoding the models for different environments.
This maximizes the investment in the analytics through reuse of
the analytical assets across environments and reduces risks by
eliminating the need for manual recoding and subsequent
revalidation: develop once, deploy many times.
Forward-thinking organizations are finding new ways to be
more efficient and drive better automated decisions. SAS Decision
Manager provides the features that organizations need for
faster and easier model deployment into production situations.
Integration Eases Model Deployment,
Monitoring and Management
While this paper focuses on the data mining and analytical
discovery process, you can’t really end a conversation about
data mining and machine learning for business applications
without touching on what happens after the predictive models
are built and the champion model chosen. So, what does
happen? You move on to the deployment phase (see Figure 2).
After the champion model has been selected, it needs to
be implemented into the right production environment.
Organizations use predictive models in different ways. For
example, they might be used to select customers for marketing
campaigns by running a batch-scoring process and providing
the selected customers as a list to marketing. An increasing
number of organizations are looking into more integrated and
automated processes to make the results of predictive models
available for operational decision making. Rather than having
the scoring process run in batch, they would like to have the
model provide on-demand answers as part of a business appli-
cation. Organizations may also want real-time answers from
streaming data (e.g., for automated fraud detection or predic-
Figure 5: SAS Decision Manager helps expedite the model deployment process. It integrates model development automation
with SAS Factory Miner and accelerates common manual tasks, like the definition of business rules and automatic generation of
Today, more organizations are recognizing the value of
predictive analytics results. And that’s good because if you’re
collecting and storing data, you should be using it to gain
insights that lead to competitive advantage.
This is especially true if your organization is paying people to
create analytical models! But the trick has always been getting
all the different pieces and parts moving together in order to
extract the maximum value from all your data. SAS offers a
complete analytical-lifecycle process that helps organizations go
from data to decisions on a very large scale, in a very reliable
It starts with data access and preparation (data volumes don’t
matter), moves through the process of data discovery and analyt-
ical modeling to produce predictive insights, and goes on to the
deployment and management of results – all in an integrated
While this paper introduced all phases of the analytical life cycle,
its main focus was on the discovery portion. And at SAS, discovery
means using predictive analytics to quickly and easily find new
and reliable insights from data. With industry-recognized data
mining software like SAS Enterprise Miner, the new SAS Factory
Miner solution, in-memory technologies and enterprise model
management capabilities, organizations are able to tackle any
big data analytics problem.
• SAS Factory Miner provides an automated, web-based
solution for building and retraining predictive models across
multiple segments. It boosts productivity by enabling
modelers to quickly and easily test many approaches simul-
taneously using machine learning and statistical algorithms.
• In situations where automated modeling doesn’t work,
SAS Enterprise Miner can be used to handcraft customized,
strategic advanced predictive models.
• Distributed in-memory computing keeps processing moving
at maximum speeds.
• SAS Decision Manager streamlines analytical model deploy-
ment – all from a single interface.
These solutions streamline the data discovery/data mining
process, enabling you to create highly accurate predictive and
descriptive models based on data analysis from across your
to find out more about
our data mining and data discovery solutions.
SAS Data Mining Community
, where users
and SAS employees share tips and other information.
For a complete overview of the entire analytical life cycle,
Manage the Analytical Life Cycle for Continuous
To learn more about the deployment phase, read
From Data to Decision: How SAS® Decision Manager
Automates Operational Decisions
To contact your local SAS office, please visit:
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. Copyright © 2016, SAS Institute Inc. All rights reserved.
Dostları ilə paylaş: