Fraud detection? Not so elementary, my dear.
April 16, 2014 Leave a comment
Radha Gowda, Technical Marketing, IBM Analytics
Did you know that fraud and financial crime has been estimated at over $3.5 trillion annually1? Identity theft alone cost Americans over $24 billion i.e. $10 billion more than all other property crimes2? And, 70% of all companies have experienced some type of fraud3?
While monetary loss due to fraud is significant, the loss of reputation and trust can be even more devastating. In fact, according to a 2011 study by Ponemon Institute, organizations lose an average of $332 million in brand value in the year following a data breach. Unfortunately, fraud continues to accelerate due to advances in technology, organizational silos, lower risks of getting caught, weak penalties, and economic conditions. In this era of big data, fraud detection needs to go beyond traditional data sources i.e. not just transaction and application data, but also machine, social, and geospatial data for greater correlation and actionable insights. The only way you can sift through vast amount of structured and unstructured data and keep up with the evolving complexity of fraud is through smarter application of analytics to identify patterns,construct fraud models, and conduct real-time detection of fraudulent activity.
IBM Watson Foundation portfolio for end-to-end big data and analytics needs
While IBM has an impressive array of offerings addressing all your big data and analytical needs, our focus here is on how DB2 solutions can help you develop and test fraud models, score customers for fraud risk, and conduct rapid, near-real-time analytics to detect potential fraud. You have the flexibility to choose the type of solution that best fits your needs – select software solutions to take advantage of your existing infrastructure or choose expert-integrated appliance-based solutions for simplified experience and fast time to value.
Highly available and scalable operational systems for reliable transaction data
DB2 for Linux, UNIX and Windows software is optimized to deliver industry-leading performance across multiple workloads – transactional, analytic and operational analytic – while lowering administration, storage, development, and server costs. DB2 pureScale, with its cluster based, shared disk architecture, provides application transparent scalability beyond 100 nodes, helps achieve failover between two nodes in seconds, and offers business continuity with built-in disaster recovery over distances of a thousand kilometers.
IBM PureData System for Transactions, powered by DB2, is an expert integrated server, storage, network, and tools selected and tuned specifically for the demands of high-availability , high-throughput transactional processing—so you do not have to research, purchase, install, configure and tune the different pieces to work together. With its pre-configured topology and database patterns, you can set up high availability cluster instances and database nodes to meet your specific needs and deploy the same day rather than spend weeks or months. As your business grows, you can add new databases in minutes and manage the whole system using its intuitive system management console.
Analytics for fraud detection
DB2 Warehouse Analytics DB2 advanced editions offer capabilities such as online analytical processing (OLAP), continuous data ingest, data mining, and text analytics that are well-suited for real-time enterprise analytics and can help you extract structured information out of previously untapped business text. Its business value in enabling fraud detection is immense.
IBM PureData System for Operational Analytics, powered by DB2, helps you deliver near-real-time insights with continuous data ingest and immediate data analysis. It is reliable, scalable, and optimized to handle 1,000s of concurrent operational queries with outstanding performance. You can apply fraud models to identify suspicious transactions while they are in progress, not hours later. This can apply across any industry segment, including financial services, health care, insurance, retail, manufacturing, and government services. PureData System for Operational Analytics helps with not just real-time fraud detection, but also cross-sell or up-sell offers/services identifying customer preferences, anticipating their behavior, and predicting the optimum offer/server in real-time.
DB2 with BLU Acceleration, available in advanced DB2 editions, uses advanced in-memory columnar technologies to help you analyze data and generate new insights in seconds instead of days. It can provide performance improvements ranging from 10x to 25x and beyond, with some queries achieving 1,000 times improvement4, for analytical queries with minimal tuning. DB2 with BLU Acceleration is extremely simple to deploy and provides good out-of-the-box performance for analytic workloads. From a DBA’s perspective, you simply create table, load and go. There are no secondary objects, such as indexes or MQTs that need to be created to improve query performance.
DB2 with BLU Acceleration can handle terabytes of data to help you conduct customer scoring across your entire customer data set, develop and test fraud models that explore a full range of variables based on all available data. Sometimes creating a fraud model may involve looking at 100s of terabytes of data, where IBM® PureData™ System for Analytics would fare better. Once a fraud model is created, you can use DB2 with BLU Acceleration to apply fraud model to every transaction that comes in for speed of thought insight.
IBM Cognos® BI DB2 advanced editions come with 5 user licenses for Cognos BI, which enable users to access and analyze the information consumers need to make the decisions that lead to better business outcomes. Cognos BI with Dynamic Cubes, in-memory accelerator for dimensional analysis,enables high-speed interactive analysis and reporting over terabytes of data. DB2 with BLU acceleration integrated with Cognos BI with Dynamic Cubes offers you a fast-on-fast performance for all your BI needs.
With the array of critical challenges facing financial institutions today, smarter are the ones that successfully protect their core asset – data. IBM data management solutions help you integrate information, generate new insights to detect and mitigate fraud. We invite you to explore and experience DB2 and the rest of Watson foundation offerings made with IBM.
Stay tuned for the second part of this blog that will explore the product features in detail.
4 Based on internal IBM tests of analytic workloads comparing queries accessing row-based tables on DB2 10.1 vs. columnar tables on DB2 10.5. Results not typical. Individual results will vary depending on individual workloads, configurations and conditions, including size and content of the table, and number of elements being queried from a given table.
Follow Radha on Twitter @radgo1