The Data Mining Group, a vendor-led consortium of companies and organizations developing standards for statistical and data mining models, has released v 4.2 of the Predictive Model Markup Language (PMML), an XML interchange format for statistical and data mining models. According to the Data Mining Group, PMML makes it straightforward to develop a model on one system using one application and deploy the model on another system using another application.
"As a standard, PMML provides the glue to unify data science and operational IT. With one common process and standard, PMML is the missing piece for Big Data initiatives to enable rapid deployment of data mining models," said Alex Guazzelli, vice president of Analytics for Data Mining Group member Zementis, in a statement. "Broad vendor support and rapid customer adoption demonstrates that PMML delivers on its promise to reduce cost, complexity and risk of predictive analytics."
Some of the elements that are new to PMML v4.2 include:
- Improved support for post-processing, model types and model elements
- A completely new element for text mining
- Scorecards that offer the ability to compute points based on expressions
- New built-in functions, including "matches" and "replace" for the use of regular expressions
PMML was introduced in late 2010. It is supported by over 20 vendors and organizations, including IBM, MicroStrategy, SAS, Experian, Pervasive Software, Zementis, Equifax, FICO, KNIME, NASA, Open Data Group, Rapid-I, Togaware and Visa.