You are here
Architected for Big Data Analytics in R
Scales Analytics Beyond the Reach of Legacy Tools
Statistical analysis of big data brings seemingly limitless opportunities to improve business results by predicting customer behavior, preventing fraud and waste, predicting maintenance events, reducing cost, predicting medical outcomes and hundreds of others.
As many open source R, SAS, SPSS and users know, big data + predictive modeling often leads to data and computational scale that is difficult, if not impossible for existing solutions to achieve. If you’re running SAS, SPSS or open source R on desktops or servers, you’ve probably encountered one or more of the following barriers:
- Inability to process large data sets due to memory or computational limitations
- Unable to meet turnaround times for model building and/or scoring
- Insurmountable delays due to moving large volumes of data between systems
- Inability to quickly deploy predictive models into production
- Inadequate flexibility or inability of commodity based architectures to coexist with production environments
- Complex problems encountered when migrating big data analytics to larger systems
- Painfully expensive upgrades to overcome any of these
Extend the Reach of R, the De Facto Standard for Advanced Analytics
Revolution R Enterprise provides breakthrough performance, scale, portability and innovation, providing users with a Big Data Big Analytics platform based on R, the de-facto standard for modern analytics. With Revolution R Enterprise or RRE, R users leverage existing R skills while RRE provides the performance, scale, portability and innovation unattainable on other platforms.
Breakthrough Performance & Scale for Big Analytics on Big Data
Revolution R Enterprise breaks through data size and performance-related barriers easily. With Revolution R Enterprise, or RRE, users can:
- Eliminate out-of-memory conditions
- Analyze data without moving it
- Build models using large data sets
- Build accurate models using ensemble techniques
- Deploy models into production without recoding
- Bring big data capabilities to users of SAS at comparatively low costs
- Do analytical development for big data entirely in the R language
Mix, Match and Migrate: Portability That Future-Proofs Analytics Investments
RRE 7 grows with your needs. As your data and analytic needs grow and change, you may require new platforms in order to quickly meet evolving business objectives. You must adapt, but do so without disrupting prior efforts to create effective analytics. With Revolution R Enterprise’s Write Once Deploy Anywhere (WODA) portability, easy platform-to-platform transitions help you avoid expensive analytics redevelopment when new platforms are needed.
Revolution R Enterprise runs R analytics transparently and portably to facilitate:
- Upgrading workstations or servers
- Building models on workstations for deployment to servers, clusters, EDWs or Hadoop clusters
- Expanding clustered systems and grids by adding nodes
- Performing data discovery, data prep and model building on Hadoop then deploying the same model to a production EDW without any delays
- Tracking the evolution of fast-changing platforms such as Hadoop
- Migrating freely between system types – workstations, servers, server clusters, Hadoop clusters and EDWs as dictated by analytical needs
With RRE, the analytics you build today on Windows Server can be run tomorrow on Hadoop or Teradata EDWs, changing nothing but the performance outcomes.
Advantages of a Big Data Capable Architecture
By overcoming the limitations of legacy or open source analytics platfom, Revolution R Enterprise broadens the range of analytical possibilities in these ways:
- Eliminate data movement and loading penalties
- Visualize extreme data sets as easily as small ones
- Train and validate a variety of statistical models using large samples
- Compute statistics and train models on data far larger than available memory
- Identify hundreds of clusters among large data sets
- Compute covariance and correlation at speeds heretofore unachievable
- Utilize stepwise regression techniques to reduce model complexity
- Rapidly develop hundreds of independent models for finely-clustered data
- Detect outliers quickly
- Score data quickly without recoding models
- Efficiently apply ensemble modeling techniques to improve prediction across diverse data sets
- Apply machine learning techniques to complex data
- Scale large or complex simulations well beyond current limits through parallel execution
- Construct fast, comprehensive recommendation engines
Components of A Scalable Big Data Big Analytics Architecture
Revolution R Enterprise scales R analytics to handle data sizes measured in terabytes and computational scale from one to hundreds of nodes, all without changing the way R is written or used.
Three Key Components of RRE’s Architecture are:
ScaleR: ScaleR provides algorithms optimized for parallel execution on Big Data. These workhorse algorithms are optimized for transparent distributed execution, eliminate memory limitations and scale from laptops to servers to large clustered systems. Click here learn more about ScaleR.
DistributedR: Adaptable parallel execution framework providing services including communications, storage integration and memory management to enable ScaleR algorithms to analyze vast data sets and scale from single-processor workstations to clustered systems with hundreds of servers. Click here to learn more about DistributedR.