SOLUTION OVERVIEW SAS Predictive Analytics Turning Your Data into Timely Insight for Better, Faster Decision Making Overview Is your organization overflowing with enterprise data but failing to turn it into useful, timely insights that help people make better, faster decisions? It s a common problem with potentially huge impacts. Without evidence-based answers to complex questions like these, you re forced to make educated guesses about: What products will different customers purchase and when? Which customers are leaving, and what can be done to retain them? How should prices be set to ensure profitability and competitiveness? How are maintenance schedules and operational influences affecting time-tofailure for key manufacturing components and how can they be optimized? How can you accurately predict fraudulent transactions? To get accurate answers to questions like these, you need powerful, multipurpose predictive analytic solutions that can turn your big data into useful insights. That s why SAS offers SAS Predictive Analytics. Designed to meet the needs of all types of users, these integrated analytics solutions make it easy to explore and analyze big data, uncover unknown patterns, find answers and opportunities, and generate insights for faster, better decision making. Challenges Are you facing any of these issues? Raw data comes from different sources and is stored in multiple formats, so it s not prepared and ready for analysis. Modelers, statisticians and analysts get bogged down with difficult-to-use and inadequate tools and algorithms. Business analysts struggle to understand complex relationships and interpret analytical results. Data gets replicated across departmental silos, hindering governance and access to timely results. Deploying models is manual and cumbersome, so people don t get what they need, when they need it. People rely on outdated or poorly performing models, leading to inaccurate projections and poor decisions. No one can explain why a model was chosen or how a particular score was calculated.
Maximizing the Value of Your Data The SAS Solution Are you just getting started with predictive analytics or have you been using it for years? Regardless of your level of experience, our solutions can help you take analytical decision making to the next level. With SAS Predictive Analytics, you can give people the right insights at the right time, dramatically increasing the reach and value of your data. And you can make evidence-based decisions a normal, everyday occurrence rather than something that happens occasionally or reactively. SAS Predictive Analytics includes powerful, multipurpose software that s easily configured for any line of business or group of users so they can: Discover relevant new insights with speed and flexibility. Analyze data to find useful results with confidence. Act to quickly make better decisions and drive better actions. Monitor analyses and results to verify continued relevance and accuracy. Manage a growing portfolio of predictive assets effectively. And because SAS products are integrated and modular in nature, you can start small to meet your current needs and add on new products as business demands change over time. Benefits Deliver Trusted Data That s Ready to Model and Analyze Is your data management team spending too much time on analytical data preparation and management? If so, you re not alone. These professionals not only have to set up, treat, and modify data for predictive models, but also keep models current and updated with the latest data. This is no small task if your organization is increasingly dependent on models and generating more data. What s needed is a way to quickly and routinely acquire, profile, sample, discover, aggregate, transform, augment, and load analysis-ready data even when it s big data. Challenges Analytical base table preparation is done in an ad hoc manner and often out of sync with model development and deployment requirements. Because algorithms used in predictive analytics require a very specific data structure, people are constantly adjusting the structure to meet the needs of each analysis. People spend too much time preparing data for analysis, as well as moving it between different data sources and the modeling environment. To increase model accuracy; provide greater context; or reveal patterns, sentiments and relationships, you need to incorporate big data structured and unstructured into models. Our Approach SAS addresses these challenges head on with industry-leading data integration and data management capabilities. No one can match the number of data sources that SAS allows you to access with ease or the ways you can manipulate and transform this data all within one environment. And our properly designed analytic database and flexible data integration toolset help you expedite data preparation tasks, reduce data movement, and flexibly add or remove data sets during model development. Discover, Explore and Understand Relationships in Complex Sets of Data Unless you can quickly uncover relationships in your data and share results with decision makers, your data will remain relatively useless. You also need ways to visualize data during the exploratory and model-building phases and share results with others. Challenges It s hard to create an environment optimized for analytical discovery and visual exploration by diverse users. It takes too long to explore large amounts of data, identify patterns, and find answers to questions. As data sets get bigger and more complex, it takes longer to generate and understand analytical results. People don t have the right tools to fully explore data.
Our Approach SAS data visualization tools enable you to transform data and create a wide variety of graphical displays ideal for different types of users from analysts to business decision makers. Using an interactive, point-and-click user interface with dynamically linked tables and graphics, people can quickly and easily engage in data discovery and visual exploration to understand relationships between data. Use Predictive Analytics to Fully Evaluate Data and Scenarios When you equip your people with easy access to data-driven evidence, you empower them make better decisions, every day. Big data-driven insights help people make decisions that are more precise and effective and they help your organization compete and differentiate more effectively. But to be effective, predictive models must be tailored for specific needs and objectives. So you need a wide selection of analytical methods and approaches to choose from, as well as a way to accurately compare them. Challenges Modelers and analysts have a very limited repertoire of analytical methods to choose from, as well as few options for tuning models. It s difficult to compare models and select the best model to support decision making. Modelers need better ways to share, integrate and reuse analytic assets. People can t interpret modeling results effectively because they lack proper business context. Organizations don t have a scalable, efficient infrastructure for model development and deployment. Our Approach SAS offers a diverse set of predictive analytic tools, including comprehensive, feature-rich statistical and data mining algorithms and model validation tools that provide multiple ways to assess how well models fit your data. You can also quickly combine and compare methods to solve more complex problems in a collaborative, repeatable manner. Model profiling functions help you understand how predictor variables contribute to the outcome of modeled data. And using in-memory processing, you can produce immediate insights and scale up to distributed massively parallel processing systems for rapid analysis of extremely large amounts of data. Integrate with Operational Systems and Processes to Eliminate Errors With increased competition and rapidly changing market conditions, who has time to wait for insight before they make decisions? Your models must be integrated with business rules for improved decision making. And outputs from models, such as predictive scores, must be used as inputs to business rules to derive faster, better decisions and fuel decision flows. To lock in the performance gains, you need to monitor model performance and guard against the risks associated with model degradation. Challenges As more people come to rely on analytics, it gets harder to create, deploy and manage models and prevent redundancy. Mistakes occur more frequently, and people don t receive their results quickly. Roles and responsibilities for the various steps of the analytics life cycle aren t clearly defined. Overlaps and gaps lead to inefficient, evidencebased business processes. It takes too long to deploy models and consumes too many resources. Over time, models degrade, so they are no longer as useful for recognizing and responding to changing market conditions. Our Approach To help you manage growing numbers of models and model types, SAS provides functionality and best practices to support complete lifecycle management of analytical models. You can also deploy models to score data stored inside the database to get faster time to insights. SAS supports an iterative approach for the selection, maintenance and continuous enhancement of analytical models for proactive decision making.
Take Decision Making to the Next Level with Integrated Analytic Capabilities Routines that took weeks or months to run now generate answers in seconds or minutes. With SAS, we re accurately scoring more than 100 million customers in seconds to target our marketing and service efforts. SAS has the software to handle the data and provide cuttingedge statistical, analytical, and visualization tools. No other software company in the world has that combination. Prasanna Dhore, Vice President, Global Customer Intelligence, Hewlett-Packard Capabilities As illustrated in Figure 1 (see next page), SAS Predictive Analytics is a comprehensive solution. Let s explore its various functional areas in more detail. Analytic Data Preparation Data preparation and data quality are key enablers of predictive modeling processes. Your systems, which may span multiple platforms and contain multiple big data sources, must be integrated and synchronized into a clear, coherent format. But manually transforming data into properly structured, representative modeling tables can take a great deal of time. With SAS, you can save time by using the software to: Structure data sets, transpose them and aggregate values into representative modeling tables. Optionally sample a representative set of data to speed processing time and in many cases, produce more reliable predictive models. Segment observations into groups, interactively bin variable values, filter outliers, replace missing values and derive new variable transformations. Replicate model parameters and data preparation code when scoring new data. Visualization and Exploration Data discovery and visualization should occur during each step of the modeling process so you can continually evaluate, explain and validate results. SAS Predictive Analytics enables you to interact with your data to explore relationships, spot trends, dig into areas that interest you and move in directions that you hadn t before considered before. Our solutions provide interactive, dynamic, and visually appealing graphics to help you: Quickly understand data, see important relationships and trends, and make better, faster decisions. Explore data from every angle possible and move rapidly from one visualization to another to identify relevant variables and discover the information needed to fuel the next step in analysis process. Statistical Analysis Powerful statistical analysis is the foundation of SAS Predictive Analytics. Use it to perform everything from simple descriptive statistics to complex Bayesian analyses, including variance analysis, categorical data analysis, survival analysis, regression modeling, experimental design, time series analysis, clustering and survey data analysis. For example, you can: Quantify uncertainty, make inferences and drive decisions by applying the right method to the right data.
SAS Predictive Analytics Analytics Key Functional Key Functional Areas Areas Analytical Data Preparation Create and Transform Variables Intuitive Role-based Interface Data Exploration and Data Discovery Descriptive and Predictive Algorithms Model Validation and Assessment Batch or Real-time Scoring Figure 1: SAS Predictive Analytics: Key Functional Areas Copyright 2013, SAS Institute Inc. All rights reserved. Measure how well models are performing against test and validation data sets, as well as when models need to be updated (or retired) to ensure accurate results. Predictive Modeling To build predictive models that will generalize to new data, you must have a wide selection of analysis tools at your disposal. SAS Predictive Analytics provides users with superior analytical depth and a broad set of predictive and descriptive modeling algorithms, including decision trees, bagging and boosting, linear and logistic regression, neural networks, memory-based reasoning, partial least squares, hierarchical clustering, self-organizing maps, associations, sequence and Web path analysis and more. SAS is constantly adding new and innovative algorithms to the software to enhance the stability and accuracy of predictions. Examples include stateof-the-art methods such as gradient boosting, time series data mining, market baskets, net lift modeling, ensembles, neural networks, random forests, and industry-specific methods such as rate making, credit scorecards. As a result, it s never been easier to: Develop models interactively to enable what-if scenarios, and incorporate business rules. Schedule processing in batch mode for large problems that may require periodic updates. Compare models side-by-side to see which approach produces the best fit. Use model profiling to understand how predictor variables will contribute to the outcome being modeled. Model Deployment Once a model has been validated, you often need to deploy it to score new data for implementation into an operational environment. This is not only a time-consuming process that can introduce costly errors into models especially when it entails manually rewriting or converting code but it can also delay model implementation. With SAS Predictive Analytics, you can: Automatically generate score code in SAS, C, Java and PMML. Deploy the scoring code in a variety of real-time or batch environments within SAS, on the Web or directly in databases and data warehouses (specifically, Teradata, IBM DB2, IBM Netezza, Aster Data, Pivotal, and Oracle).
How SAS Can Help Are you looking to predict customer behavior and intentions? Reduce customer churn? Minimize credit risk? Combat fraud? Reduce asset downtime? Or anticipate patients at risk? Regardless of your industry and need, SAS predictive analytics can help you create optimal data models for multiple scenarios and get the answers you need. Imagine being able to: Select, explore, transform and model large quantities of data using a variety of classical and modern analytic algorithms to identify opportunities and mitigate risks quickly and reliably. Use an intuitive data discovery and visualization environment to quickly and interactively explore data and discover new insights. Establish and share repeatable, collaborative predictive modeling processes and update models with new data to move more quickly from a discovery environment to operational analytics. Use a patented, secure model repository with a rich metadata structure and easy-to-use project templates to streamline the deployment and management of models. Use model monitoring features and version control reports to manage the entire life cycle of models from creation and deployment into production systems to retirement. Solve complex problems faster and more efficiently using high-performance analytics. Model Management and Monitoring Moving a model from a development environment to an operational environment requires continuous collaboration between your analytical teams and IT. SAS Predictive Analytics facilitates this collaboration to help ensure only the best models ones that will help improve organizational performance get deployed. Using a workflow console delivered as an easy-to-use Web based client, you can centrally manage models, automate key activities (including registering, validating, publishing, scoring and monitoring models), and track progress through each step of the modeling process. You can also: Easily collaborate and reuse models. Use a self-documenting process flow to efficiently map the entire data mining process. Set automated alerts to detect when scoring results are changing over time, which may indicate model decay. Efficiently map the entire data mining process using a self-documenting process flow diagram. Produce compliance and validation reporting increasingly required for regulatory compliance. Capture valuable best practices via a patented centralized data repository, lifecycle templates and a metadata management system. High-Performance Analytics High-performance analytics solutions give you an opportunity to derive value from big data, solve complex business problems and deliver timely insights using a reliable analytics infrastructure. With SAS High-Performance Analytics, you get scalable and reliable analytical processing that makes the best use of your existing IT infrastructure resources. It can help you solve industry-specific complex problems and perform analyses that range from data management, data visualization and exploration to model development and deployment. The key enabler of high-performance analytics is in-memory processing, which has gained lot of attention in today s marketplace. SAS In-Memory Analytics gives you concurrent access to data, no matter how big or small. The software is optimized for distributed, in-memory processing so you can run new scenarios or complex analytical computations at blazingly fast speeds.
As a result, SAS In-Memory Analytics can help you: Gain fast access to deeper insights so you can seize opportunities and mitigate threats in near-real time. Run more sophisticated queries and models using all of your data not just a subset to generate more precise insights that can improve business performance. Run analytical workloads needed to perform analyses ranging from data exploration, visualization and descriptive statistics to model building with advanced algorithms and scoring of new data. With an unwavering focus on data analytics since 1976, SAS offers a broad set of tools for predictive analytics, an architecture that supports multiple platforms, in-database analytics, in-memory analytics, and significant market presence. The Forrester Wave : Big Data Predictive Analytics Solutions, Q1 2013, Forrester Research, Inc. January 3 rd, 2013
Give People the Right Analytical Tools at the Right Time You can find relationships that aren t readily apparent when you run standard regression models. [SAS] Enterprise Miner made it so darn easy. It cut down on the amount of work spent churning through data to look for relationships. Daryl Wansink, Director of Health Economics, Blue Cross Blue Shield of North Carolina Components SAS Visual Analytics SAS Visual Analytics is an in-memory solution for data exploration and visualizing massive amounts of data very quickly. Use it to spot patterns, identify opportunities for further analysis and convey visual results via Web reports or mobile devices. You can also execute analytic correlations on billions of rows of data to identify which variables influence each other, understand how the value of a target variable is affected (if at all) by a set of different input variables, and quickly identify the presence of outliers in a specific segment of data. You ll be identifying previously hidden relationships that need further analysis and accelerating the model development process with ease. Best suited for... Organizations interested in speeding up their modeling processing by using a highly interactive data visualization tool. No matter your skill level, you can use SAS Visual Analytics to quickly identify areas that warrant further exploration and analysis, rather than resorting to trial and error. Business analysts can also use it to try different methods to identify the data that will determine an outcome with a reasonable level of certainty. SAS Enterprise Miner SAS Enterprise Miner streamlines the data mining process so your analysts and modelers can create highly accurate predictive and descriptive models based on vast amounts of data. The user interface is designed to support the iterative and interactive nature of the complete data mining process. Because the software is customizable and extensible, you can readily integrate your code and build new applications for redistribution. And using multithreaded algorithms that take advantage of either in-memory analytics, in-database analytics or grid computing processing options, you can build models using big data. These algorithms dramatically reduce model execution time and use hardware resources more efficiently. The software also provides complete, optimized scoring code in SAS, C, Java and PMML, which you can use to score models deployed in both SAS and other environments. Best suited for... SAS Enterprise Miner is suitable for organizations that need to deploy distributable and collaborative data mining capabilities throughout the enterprise. These organizations typically are focused on repeatable data mining that is often administered through multiple user-workgroup collaborations. Those who need to score new data in batch and real time should also choose SAS Enterprise Miner. SAS Enterprise Miner for Desktop SAS Enterprise Miner for Desktop brings the power of data mining to analysts, modelers and statisticians in small and midsize firms or those who work independently in departments within large organizations. SAS Enterprise Miner for Desktop runs entirely
within the confines of a Windows PC. It provides the same user interface and features found in SAS Enterprise Miner (i.e., the client/server version), enabling a direct upgrade path. It also provides optimized SAS score codes for putting the model into production. Best suited for... Small and midsize businesses that need a single desktop implementation of SAS Enterprise Miner to support only a few users. SAS Model Manager SAS Model Manager streamlines the tedious and often error-prone steps of registering, validating, publishing, deploying and monitoring analytical models. Using a Web-based workflow client, you can facilitate collaboration between analytics, model validation and IT team members and track progress, streamlining the management and deployment of analytical models into production. The software also provides an integrated environment for tracking and monitoring the performance of your models so you can decide whether to update, retire or create new ones. SAS Model Manager also includes unique compliance and validation reporting capabilities, which are highly sought after by those facing heightened regulatory requirements. Best suited for... Any organization that needs to manage and monitor large collections of predictive and descriptive models developed with SAS Enterprise Miner, SAS/STAT (for linear models only) or SAS/ETS (for select models). SAS Scoring Accelerators With SAS Scoring Accelerators, you can publish scoring models created in SAS Enterprise Miner, SAS/STAT linear models or SAS/ETS (for select models) into database-specific functions to score new data inside the database. SAS Scoring Accelerator automates model-scoring processes, reduces data movement and uses parallel processing capability offered by the database to deliver quick, timely results. Best suited for... Organizations that use Teradata, IBM Netezza, IBM DB2, Aster Data, Pivotal, Oracle, Hadoop (Cloudera Distribution) databases or data storage systems and want to extend the value of these IT investments. In-database scoring is a best practice for high-performance analytic deployments, and enterprises can choose the scoring accelerator engine best suited to deliver faster insights. SAS Analytic Accelerator for Teradata SAS Analytic Accelerator for Teradata enables in-database processing of a core set of data discovery and summarization, statistical and analytic functions within a Teradata database. You can prepare representative samples, accumulate time series data and build predictive and descriptive models inside the database to avoid large-scale movements of data, reduce processing time, maintain data consistency, apply your existing SAS skills and promote better data governance. Best suited for Organizations that have invested in Teradata and want to conduct data discovery and analytical model development steps inside the database. SAS High-Performance Data Mining SAS High-Performance Data Mining allows you to develop predictive models using big data, not just a subset, and select from thousands of variables to produce more accurate, timely insights. A highly scalable, distributed, in-memory processing architecture allows you to process models in minutes or seconds, perform frequent modeling iterations and use sophisticated analytics to get answers to questions you never thought of or had time to ask. This solution is available on Teradata, Pivotal or Oracle data compute appliances, as well as on commodity hardware using Apache Hadoop (Cloudera Distribution). Best suited for Organizations interested in deriving insights at breakthrough speeds for high-value and time-sensitive decisions. It offers more precision when seeking answers to questions by allowing users to analyze big data, incorporate unstructured data, perform more iterations and use complex analytical techniques. Modelers, statisticians and data miners can also use it as a dedicated and scalable in-memory analytics infrastructure to execute more ideas and test multiple scenarios using all of their data.
SAS Predictive Analytics: Empowering Organizations Around the World The SAS Difference SAS Predictive Analytics supports an integrated predictive modeling process that helps you solve complex problems, exploit your data assets and drive better performance. You can: Move from pockets of analytic excellence to the pervasive use of evidence-based decision making. Facilitate continuous enhancement, refinement and maintenance of the analytical models that drive your decision-making processes. Grow and standardize on a common platform with multiple entry points for predictive analytics and data mining. Build models that generalize well and produce superior outcomes by using flexible data preparation and data management capabilities. Take advantage of a rich, interactive visualization and data exploration environment to quickly identify the best opportunities. Harness the power of comprehensive, feature-rich statistical and data mining tools and cross-domain model governance capabilities. Achieve better response time and faster insights with in-memory analytics. Streamline the exchange of assets via an integrated metadata repository that documents each step in the modeling process. Credit card, banking and financial services companies use SAS predictive analytics to detect and reduce fraud, measure credit risk, maximize cross-sell/up-sell opportunities, retain customers and optimize marketing campaigns. Insurance companies use SAS predictive analytics for determining insurance premium rates, detecting claims fraud, optimizing claims processes, retaining customers, improving profitability and optimizing marketing campaigns. Governments and the public sector use SAS predictive analytics to improve service and performance; detect and prevent fraud, improper payments and the misuse of funds and taxpayer dollars; and detect criminal activities and patterns. Telecommunication companies use SAS predictive analytics for segmenting customers, reducing customer churn, retaining profitable customers and developing effective cross-sell/ up-sell campaigns. Health care providers use SAS Analytics to predict the effectiveness of new procedures, medical tests and medications, and improve services or outcomes by providing safe and effective patient care. Health insurers use SAS predictive analytics for detecting and handling insurance claims fraud, identifying which patients are most at risk of chronic diseases and knowing which interventions make the most medical and financial sense. Manufacturers use SAS predictive analytics to identify factors leading to reduced quality and production failures, as well as to optimize parts, service resources and distribution. Media and entertainment companies use SAS predictive analytics to deepen their insight into audiences by identifying influencing attributes, trends, drivers and desires across properties, and scoring visitors to determine appropriate audience segments and behavior value. Oil, gas and utility companies use SAS predictive analytics to get a unified view of facility assets, diagnose problems, predict failures, mitigate safety and reliability risks, and improve performance. Retailers use SAS predictive analytics to assess the effectiveness of promotional events and campaigns, predict which offers are most appropriate for consumers, and determine which products to stock where and how to build brand loyalty.
SAS is our standard for predictive analytics. As we continually look for ways to drive revenue for the organization, SAS helps us identify how much to invest and where. SAS gives us the innovative muscle to support the aggressive goals of Kelley Blue Book. Dan Ingle, Vice President of Analytic Insights Technology, Kelley Blue Book 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 65,000 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. sas.com
SAS Institute Inc. World Headquarters +1 919 677 8000 To contact your local SAS office, please visit: sas.com/offices 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 2013, SAS Institute Inc. All rights reserved.103121_s95554.1213