Thursday, January 24, 2013

R PMML Support: Data Transformations


R and PMML Export 
  
R is becoming the tool of choice for many data scientists. It is no wonder that many commercial and open-source statistical tools are also embracing R.

Predictive Models

A set of robust predictive analytic techniques is but one set of tools available to data scientists in R. Another important set is the ability to export PMML for a host of predictive models. 

By using the pmml package (version 1.2.33 or higher), users can export PMML from R for:
  • Random Forest Models
  • Neural Networks
  • Clustering Models
  • Cox Regression Models
  • Linear and Logistic Regression Models
  • Support Vector Machines
  • Association Rules
  • Generalized Linear Models
  • Random Survival Forest Models

Data Transformations

And now, another R package extends this functionality by providing PMML export for data transformations. The new pmmlTransformations package has just made its way to CRAN (the Comprehensive R Archive Network). 

Want to apply a Z-scoring normalization procedure to your continuous input variables before presenting them to a neural network? No problem. Use the pmmlTransformations package in conjunction with the pmml package (version 1.2.33 or higher) to export the entire process (pre-processing + model) into a PMML file. 

To look at the package's documentation in CRAN, click HERE.

Agile Predictive Analytics Deployment

Once represented as a PMML file, a predictive solution (data transformations + model) can be readily moved into the operational environment where it can be put to work immediately. That's the promise of PMML.

Zementis offers a host of products for the agile deployment and execution of your PMML-based solutions. Our ADAPA and UPPI scoring engines are available for:
  • Hadoop: Datameer and Hadoop/Hive
  • In-database: EMC Greenplum, IBM Netezza, SAP Sybase IQ, Teradata, and Teradata Aster
  • Cloud: Amazon EC2 and IBM SmartCloud Enterprise
  • On-site: On your own servers
Real-time or Big Data requirements? Zementis has you covered.

Contact us today for more information or to schedule a presentation/demo.

No comments:

Post a Comment

Welcome to the World of Predictive Analytics!

© Predictive Analytics by Zementis, Inc. - All Rights Reserved.





Copyright © 2009 Zementis Incorporated. All rights reserved.

Privacy - Terms Of Use - Contact Us