Building on BI with Predictive Analytics

Information Builders strengthens its WebFOCUS platform, delivering on cost-effectiveness and demands for analytics, coming closer to a platform. By integrating a leading open source engine, they will extend the value and accessibility of predictive analytics in production BI applications across the enterprise.

Today's transactional systems capture and store large volumes of multi-dimensional data. With hundreds of variables, the possibilities to slice and dice can quickly become as overwhelming as an Excel report or PivotTable with hundreds of columns and thousands of rows. Data mining and statistics are more efficient methods to analyze such data sets because analysts can quickly understand which of the many variables are important. Furthermore, they can use the historical data to generate predictions about the future for anything from credit risk and crime occurrence to insurance and medical risks.

"Data mining has been largely excluded from mainstream BI applications, managed as one-off projects where a specialist conducts the analysis and reports the findings," says Information Builders CEO and founder Gerald Cohen. "Very rarely has prediction become part of the daily operations of field workers. Information Builders believes that every decision-maker at any level in the organization can and should benefit from access to predictive modeling applications. Embedding them within WebFOCUS allows us to extend the use of these models to users on all platforms in many, many different kinds of applications."

Expected at year end, WebFOCUS RStat is a statistical modeling workbench embedded in Developer Studio that enables scoring of new data on every platform and database, even in real time. RStat allows data miners and BI developers to work collaboratively with the same tools to access, manipulate, or transform data; develop predictive models; and create and deploy Web applications along with associated reports to any worker within their organization. By incorporating RStat, the widely used open source statistical library product among the world's universities, Information Builders dramatically reduces the customers' cost of integrating these robust technologies.

Police officers, for example, planning without predictive tools rely on intuition, but predictive models allow them to benchmark their intuition and experience against past data in the system. They can verify their crime expectations against predictions based on past crime data to gain a more accurate view and deploy resources effectively. If a call center operator answers a call and that customer's lifetime score is automatically generated from a scoring application, the operator can tailor the conversation accordingly. Similarly, an underwriter of specialty insurance policies empowered with analytics can benchmark the cost of a policy against the data and price a policy based on the most important variables and the entire insurance history for such policies.

Each of these field workers benefits from modeling work done by specialists. By enabling different tiers of specialists to work on the same platform within a robust BI system, Information Builders gives customers unified applications that use expert knowledge and have immediate relevance and value in the field. An integrated environment like WebFOCUS RStat eliminates the need to work in different and often incompatible environments and eases the formation and effectiveness of BI competency centers. The product can be deployed on any platform so any organization can take advantage of Information Builders' entire data access and data transformation capabilities to power statistical applications.

Highlights of WebFOCUS RStat will include:

  • Integrated, intuitive user interface in Developer Studio to access the most widely used statistical and data mining models: decision trees, neural networks, linear and logistic regressions, random forests, support vector machine, boosting, association rules, K-means, and hierarchical clustering
  • Ability in Developer Studio for modelers to prepare the data, i.e., extract it from any data source, manipulate and transform fields to prepare data for analysis, and do the modeling
  • Access to a simple Web form so that end users can, with the click of a button, generate scores and predictions without having to know anything about data mining and modeling
  • Predicted values and scores can be incorporated in any WebFOCUS report so that users can benchmark their intuitions and expert opinions to ensure that they have taken all factors into consideration when making decisions
  • RStat will be extensible via scripts and open source to leverage the numerous packages developed by industry experts and scientists to perform highly complex analyses and rapidly deploy applications

Read more…