This Week’s Big Data Big Seven
This week we’ve been planted in San Jose for the GPU Technology Conference, following developments here, including news of virtualized GPUs and the unveiling of NVIDIA’s Kepler architecture for high-end computing.
To wrap up the week ending today, May 18, in news, we will take a brief look at how IBM Watson is being put to new use, how Caltech sought to address its storage limitations, see how insurance fraud is being targeted, and relay an interesting use case from an unexpected source on the retail side.
Let’s get started…
Putting Watson to Work
While most people outside of computing fields are familiar with IBM’s Watson supercomputer from its appearance on Jeopardy, it wasn’t just built to entertain the masses on quiz shows.
The technology behind Watson was designed to address both data and computationally-intensive workloads in critical fields and has already been tapped for roles in healthcare, among other areas.
This week the winners of a new round of thought leadership about making use of Watson were announced.
Twenty-five MBA and master of science students with concentrations ranging from marketing and business consulting to finance and entrepreneurship competed in teams submitting seven proposals outlining how Watson’s technology could be applied to solve complex challenges in the transportation, energy, retail and public sector industries.
Three winning ideas were selected by a panel of judges comprised of faculty, regional business leaders and IBM executives. Team evaluations were based on the ability of the students to clearly articulate the business case including market research, tactical planning and feasibility while exhibiting an understanding of how to harness big data for strategic outcomes.
The top award went to a team that tapped Watson’s ability to look at unstructured and structured information could more accurately identify weather patterns and help improve response times.
The runners-up used Watson technology to help energy companies improve the understanding of the environmental impacts of oil and gas discovery and production, as well as to better gauge the regulatory and safety information to reduce accidents.
Another team, this one in third place, devised an approach employing Watson’s technology to quickly analyze massive amounts of unstructured information in order to enhance security, reduce wait times and improve the travel experience in airports while taking the guesswork out of the customs process.
Caltech Looks to Parallel Storage
This week Panasas announced that Caltech’s Center for Advanced Computing Research (CACR) has installed Panasas ActiveStor 11 to deliver high performance parallel storage as part of its newly upgraded high performance computing (HPC) facilities for advanced computational science and engineering.
CACR operates large-scale computing facilities and provides support services for numerous campus research groups that require reliable I/O, including the aeronautics, applied mathematics, astronomy, biology, engineering, geophysics, materials science, and physics departments.
Because CACR previously experienced non-critical data loss with its legacy storage systems, and endured the use of problematic system administration tools, it was important to ensure that manageability, reliability, and stability of the parallel file system and its underlying storage hardware was best-in-class.
“We needed a high performance storage solution that was big enough and fast enough for our I/O demands, and that would not get in the way of our research. It was key to be able to take file system usability and customer support as a given,” said Sharon Brunett, senior scientist at CACR who was tasked with the overall Panasas ActiveStor selection and installation. Brunett went on to note that this choice eliminated many file system administration and system management hassles, as well as user complaints about lackluster performance and application response times.
ActiveStor appliances aim to eliminate the bottlenecks found in traditional NAS systems, accelerating application I/O performance by enabling HPC cluster nodes to directly access a single, scalable file system in parallel. Users add individual blade chassis or entire racks to non-disruptively scale the capacity and performance of the file system as storage requirements grow.
Predictive Analytics Nabs Insurance Fraudsters
Santam, South Africa’s leading short term insurance company, has saved $2.4 million on fraudulent claims in the first four months by leveraging predictive analytics from IBM. The software has enhanced Santam’s fraud detection capabilities and also enabled faster payouts for legitimate claims.
The claims division developed a new operating model for processing claims, depending on varying risk levels. IBM’s predictive analytics software has enabled Santam to automatically assess if there is any fraud risk associated with incoming claims and allows the insurer to distribute claims to the appropriate processing channel for immediate settlement or further investigation, which optimizes operational efficiency.
With the enhanced claims segmentation, Santam is also able to reduce the number of claims that need to be assessed by mobile operatives visiting the customer or claim site, resulting in further considerable cost savings for the company.
Speed of claims handling is an important differentiator for the company. Before using IBM analytics, it took at least three days to settle claims. Now, Santam is able to settle legitimate claims within an hour allowing the insurer to significantly improve customer service.
According to Anesh Govender, Head of Finance, Reporting and Salvage at Santam, IBM and their business partner OLRAC-SPSolutions have helped the insurance company to build a solution that transformed their claims processing methodology in terms of speed and efficiency and offered a new view into fraudulent activity. He says the solution has “delivered a full return on investment and also helped uncover a motor insurance fraud syndicate in less than 30 days after the system went live.”
NEXT — Serving up Data Analytics >
1010 Data Serves Up Analytics
SONIC, the nation’s largest chain of drive-in restaurant selected 1010data’s Internet-based Big Data Warehouse to analyze billions of records to enable critical business decisions ranging from operations to marketing.
At the core 1010data’s approach is a flexible data warehouse and cloud-based based platform that SONIC officials say provided the speed and flexibility to analyze and report huge amounts of business data across the chain’s over 3500 locations.
“1010data lets our analysts use all available data in an intuitive, interactive way,” said Craig Miller, CIO of SONIC. “This makes analytics a flexible, exploratory process that encourages discovery — allowing us to make data analytics a part of every decision we make.”
According to Tim Negris, VP of Marketing of 1010data, “Quick Service Retail is a zero sum game where fast data analytics is the key to competitiveness. Making better, faster decisions than the competition on things like customer segmentation, price optimization, staffing, real estate selection, and promotion management requires timely, accurate analytics. In a business where minutes mean millions, finding problems and discovering opportunities faster than the other guy makes all the difference. SONIC has always been a product and process innovator and 1010data is helping them take it to the next level.”
NEXT – Opera Sings on SAP HANA >
Opera Sings on SAP HANA
This week Opera Solutions announced that it melded its Signal Hub technologies and related applications with the SAP HANA platform to support predictive analytics delivered in real or near-real time.
The company’s Signal Hub technologies use advanced machine learning techniques and proprietary ensemble approaches combining hundreds of models to extract predictive patterns, or “signals,” from big data flows.
The signals from big data are used in its applications that bring an ongoing stream of machine-generated recommended actions. The company says that SAP HANA’s capabilities can streamline signal extraction and processing for Signal Hubs’ computationally intense activities. In addition, the platform’s Predictive Analytics Library includes advanced algorithms such as K-Means clustering, which run natively on the platform to enable real-time Signal generation.
According to Arnab Gupta, CEO of Opera Solutions, by allowing us to keep both time series history and complex event processing in memory, users can reduce speed-to-answer and simplify their processing environment. He adds that this moves his company’s predictive analytics closer to real-time.
Opera Solutions has been exploring use cases involving SAP HANA and Signal Hub technologies for global financial services firm Morgan Stanley, where Opera Solutions has previously installed its Wealth Management Signal Hub powering a range of investment management applications that are in use by the firm’s 15,000+ Financial Advisors.
Informatica Eyes Insurance Industry with Updates
Data integration software company, Informatica pinned two new data integration capabilities to its offerings within Informatica 9.5 that are designed to appeal to the particular big data needs of those in the insurance industry.
At the heart of the announcement are two tools: XMap and HParser. Working alone or in concert, the improved capabilities enable insurance organizations to more effectively address the opportunities and challenges around compliance—namely issues related to the ACORD insurance standards and big data.
In a recent customer-hosted benchmark, XMap and HParser were used in the processing of three gigabytes of proprietary insurance XML on Amazon Elastic MapReduce (Amazon EMR). According to Informatica, this entire round of XML processing took only 50 seconds on a 16-node Amazon EMR cluster.
As Ash Kulkarni from Informatica noted, “XMap enables the complex ACORD XML schemas to be mapped with ease and HParser efficiently executes XML transformations natively on Hadoop clusters for cost-effective, high-performance processing of big insurance data.”
From Informatica’s point of view, XMap and HParser empower insurance organizations to maximize their return on big data by lowering the costs of mapping, transforming and processing big insurance data. Their pitch to insurance companies is that they can increase the value of the data through improved business access to information, more efficient data-driven business processes and the ability to engage in big data analytics.
Next – Powering Big Real Estate >
Powering Big Real Estate
RE/MAX, which oversees a network of nearly 90,000 real estate agents in more than 80 countries with approximately 6,300 office locations has a great deal of data to contend with across a number of disparate business units. To contend with the complexity, the company announced it looked to enterprise analytics company, Microstrategy.
RE/MAX will use MicroStrategy to provide stakeholders with the ability to address everything from finance, operations, membership and sales data. In its statement about the selection, the real estate giant cited the platform’s architecture and product capabilities, its ability to be deployed from almost any location or device, and the overall performance.
“MicroStrategy received the highest marks in all phases of our competitive business intelligence evaluation process and had the most compelling total cost of ownership,” said Tim Drouillard, Vice President, Information Technology at RE/MAX, LLC. “We look forward to giving our management team and information workers a superior BI foundation to report on all facets of the business in a timely and intelligent manner.”
“We are delighted that RE/MAX has chosen MicroStrategy to provide its master franchise operations with new BI capabilities and faster reporting,” said Sanju Bansal, MicroStrategy Chief Operating Officer. “MicroStrategy’s interactive dashboards and superior analytics empower leading companies like RE/MAX to easily interact with data, so that decision makers spend less time looking for answers and more time driving strategic decisions.”