You are currently browsing the tag archive for the ‘Predictive’ tag.
In our benchmark research in predictive analytics we’ve uncovered some intriguing tools for taking advantage of big data in the enterprise. Revolution Analytics, which we analyzed earlier this year, this month introduced its 6.0 release. Revolution extends the open source statistical programming language R with capabilities you would expect out of enterprise software. The company has grown substantially over the last several years and has an impressive list of more than a hundred customers in both the private and public sectors. Revolution partners with database and data integration providers such as Talend and Informatica and business intelligence providers who want to connect to more advanced level of analytics. Revolution can operate across a range of big data architectures, including Hadoop, working with Cloudera and IBM as well as data warehouse appliances such as IBM Netezza and Teradata. This is a smart move, since predictive analytics is the second most important unavailable capability cited by big data deployments in to our benchmark research.
With version 6, Revolution Analytics can now operate across grids of computing technology supporting Platform Computing’s LSF (Load Sharing Facility) clusters for analytic jobs operating across Linux servers. Revolution also can be managed within Microsoft HPC Server management tools to operate in the Azure Cloud to get more elasticity in compute power. Users can prototype locally then deploy into test and production environments. Revolution supports generalized linear models (GLM) to help with the design and deployability of predictive analytics. To support the largest obstacle reported by organizations in our benchmark research, difficulty integrating into information architectures, Revolution supports non-XDF (eXtensible Data Format) data sources, with direct support of ASCII and ODBC but also SAS and SPSS without having to install those platforms. If needed, data can be transformed to XDF format for further analytics in Revolution. For tighter integration with Hadoop, Revolution built RevoConnectR, which allows customers not just to integrate with HDFS but also import tables via Apache HBASE and write map-reduce tasks from within R. All of these advancements in version 6 help broaden the potential for not just the design and modeling aspect of Revolution but also for supporting deployment of the models in business processes.
Achieving a competitive advantage, identifying new revenue opportunities and increasing profitability are the top benefits we found in our benchmark on the value of predictive analytics. Predictive analytics help businesses to be more intelligent about the decisions they make every day. Revolution, with its latest release, provides more flexibility and openness to its technology while becoming more integrated with the range of platform and information architectures that IT organizations operate with today and that they will use in the future.
Mark Smith – CEO & Chief Research Officer
At first I thought 1010data just developed a faster data warehouse technology to be used with business intelligence tools. After a recent briefing, however, I learned that it provides much more than a data warehouse; the product is a large-scale analytics server that empowers business analysts to work on big data, conducting for, example, market basket analysis or correlations of customer and product information. The software lets organizations retain and analyze more data and increase the speed of analysis, which our benchmark research on big data found to be the top two benefits of the technology for more than 70 percent of organizations.
I had a chance to review the technology and its capabilities and for a person that loves business analytics and ways technology can help analysts, I was quite impressed. 1010data provides big data and business analytics in the cloud and in a hosted environment, allowing companies to avoid having IT implement and maintain the technology or the data. This hosted and software-as-a-service approach is demanded by more than half of organizations today according to our big-data benchmark research. It has a Web-based interface for conducting analytics against the customer organization’s existing enterprise data and external data that can be acquired through 1010data. There is also a simple interface through which users can develop data sets quickly for analysis.
For analysts, 1010data provides capabilities to perform granular statistical and predictive analytic routines that can be extended through its own language and interface. Within minutes users can segment data for further analytics. The vendor calls this the Trillion-Row Spreadsheet. Our benchmark research in business analytics found that analysts spend more than two-thirds of the analytic process in data-related activities; 1010data can reduce this preparation significantly. The company also provides a Microsoft Excel add-in that lets users adopt a spreadsheet as the environment for reviewing data and for creating charts or presentations. Its Micro-Segmentation Wizard helps users target segments in its analytics. All of these capabilities let 1010data deliver on business analytics in a way that traditional business intelligence is not able to operate efficiently against big data and apply analytics to get results within seconds.
You can also access 1010data from business intelligence tools such as Information Builders’ WebFOCUS. However, Information Builders needs to strengthen its interface to 1010data if it wants its customer to fully enjoy its performance and scalability. The use of business intelligence helps get the analytics and resulting metrics out to business users in the organization who need the metrics to take actions and make decisions and can be included within dashboards, reports or other analytic related deployments.
For IT, 1010data loads customer data into its secure, managed environment through an interface called Power Loader. It also provides an API for organizations that want to integrate their existing processing with the 1010data platform, as well as a script-level interface called TenDo. These capabilities can be integrated with IT level processes related to data to ensure that the data related tasks like data integration and quality are managed by IT while the analytic capabilities are provided to those who are held accountable for the use of them in business. This combination of IT working with business on areas of information management is still a point for improvement in over half of organizations according to our information management benchmark research.
While 1010data started in the financial services market, it has seen significant growth in the manufacturing and retail markets, as it can manage massive volumes of transaction data across customers and consumer interactions at stores, on the Internet and via social media. This variety of data that is typically large in size and that needs to be analyzed at the detail level is what our benchmark in big data found with customer and transactional data being the most common to manage and analyze. In 2011 it signed up new customers such as Coca-Cola, Nestle, Purina and SC Johnson.
The immediate challenge for 1010data is to get the market to understand that it offers more than a data warehouse. The platform provides big-data analytics that do not require organizations to learn new technologies but let analysts move beyond silos of spreadsheets and reporting to real analytics. It helps build the foundation to more advanced analytics like predictive analytics that our business analytics benchmark found to be very important in 37 percent of organizations and achieve the most important benefit of them found in our predictive analytics benchmark research which is achieving a competitive advantage in 68 percent of organizations. If your analysts need more power and less time worrying about the data issues in their jobs, consider 1010data especially if you are an analyst or in IT who wants to improve the value of data and working relationship with business.
Mark Smith – CEO & Chief Research Officer