SAS Introduces Visual Analytics for Big Data
In a demonstration of its new data exploration and visualization solution, SAS took pains to differentiate what it calls high-performance analytics from simple business intelligence.
In a blog post written last spring, Keith Collins, the chief technology officer of SAS, listed the company’s top five research and development priorities for 2011: software-as-a-service, data management, business visualization, process automation and high performance analytics.
A year later SAS is delivering Visual Analytics, a data exploration and visualization solution that leverages an in-memory architecture designed to scale and handle large amounts of data. In a demonstration offered during a press preview, SAS Vice President of Product Management Randy Guard put the solution through its paces with 1.15 billion rows of data.
Taking pains to differentiate Visual Analytics from products offered by competitors, Guard said it “is not a relational database relying on SQL.” Rather, he said, it was engineered from the ground up to offer the ability to perform sophisticated analytical calculations in memory.
SAS CEO Jim Goodnight also stresses this message in a video on the company’s website, saying that a SQL database in memory can’t be used for tasks such as regression analysis “because that type of computational ability isn’t built into databases and probably never will be.“
Visual Analytics leverages the SAS LASR Analytic Server and uses Hadoop (embedded Hadoop Distributed File System) as local storage at the server for fault tolerance. As the company points out, SAS LASR Analytic Server has been tested on billions of rows of data and is extremely scalable, bypassing the known column limitations of many relational database management systems.
Working with larger data sets and reducing the amount of time required for calculations opens up new business opportunities for users. Using the billion rows of data, Guard in seconds produced a correlation matrix so he could look at variables and quickly see which were highly correlated and which were not correlated at all. The value of this kind of analysis increases with the amount of data used, he said.
“Business users don’t have to limit themselves to subsets of data when doing analysis and solving problems. They cal look at all store and sku combos or all call records. Our early customers are seeing the value of this,” Guard said.
A SAS announcement includes a statement from Antoine Georges, vice president of Analytics at Virgin Mobile USA, a Sprint prepaid company, who said, "I could immediately recognize in SAS Visual Analytics the potential benefit of incorporating the massive amount of information in our call detail records into our models, but also the value of scoring and targeting our customer base in real time with the most relevant offers.”
Visual Analytics is also designed to make business analysts and other savvy business users less reliant on IT organizations. Guard showed a feature called auto charting, which automatically renders data in the visual format deemed most appropriate. If a user changes the data properties, the visual quickly changes. In addition, the user can overrule the tool and select a different visual format.
“The user creates the hierarchy on the fly,“ Guard said “There are no drill paths or cubes created in advance.”
In addition to the SAS LASR Analytic Server that calculates data in memory, the solution includes an ad hoc discovery and visualization tool called explorer, a designer tool that allows users to create custom reports and publish them to Web or mobile environments, a mobile support tool for downloading and viewing reports on mobile devices, and an administration environment used by administrators to manage users, security and data.
In the initial release of Visual Analytics, available next week, Guard said SAS is supporting Apple’s iOS. Subsequent releases will add support for the Android operating system.