Taking Advantage of Big Data with Predictive Models
90% of data in the world was created in the past two years. With staggering statistics like these, it is essential to leverage new and relevant data when making important business decisions. Big data can be useless without having the proper tools in place to model and analyze it. Predictive analytics offers a way to utilize this data in making more informed business decisions.
Predictive analytics is based upon highly rigorous mathematical and statistical models built by experienced individuals in a development environment. Exporting these models into the production environment, previously took days, or even months; causing important business decisions to remain in a limbo while technical experts recreated the model in an enterprise data warehouse. With the adoption of Predictive Model Markup Language (PMML), one of the main obstacles that slowed the implementation of analytical applications has been removed. PMML is now the standard way for an application to define statistical and data mining models. PMML models can be shared quickly and easily with any other PMML supporting application. Implementation time is dramatically reduced by automating the process of exporting the model from development to the data warehouse execution environment.
SAP Sybase IQ 16 supports PMML with the aid of Zementis Universal PMML Plug-in (UPPI). Zementis UPPI converts the PMML into Java UDFs which can be invoked using SQL. The plug-in automatically recognizes compliant predictive models that support versions of PMML standard 2.0 and up. Analysts can use a wide variety of development tools to create different types of models like Decision Trees, Neural Network models, Naïve Bayes Classifiers, Cox Regression models, and more. These models can then be effortlessly exported to the SAP Sybase IQ production database to see results almost immediately! This speedy deployment is made possible by the Zementis UPPI, thereby allowing models to be modified and quickly put back into operation to reflect the most up-to-date business conditions.
SAP Sybase IQ 16 shatters boundaries that previously existed for complex models due to resource or performance constraints. The unique PlexQ shared-everything, massively parallel processing (MPP) columnar architecture allows automatic re-balancing of workload to avoid conflict among users for resources. This means analysts in every department inside an organization can build complex, resource-intensive analytical models that can be used frequently by as many users as necessary, on massive shared data sets.
You can learn more about how Zementis UPPI works with SAP Sybase IQ here: http://www.zementis.com/docs/Universal_PMML_Plug_in_for_Sybase.pdf