PMML stands for Predictive Model Markup Language. PMML provides a way for analytic applications to describe and exchange predictive models produced by data mining and machine learning algorithms.
The standard has been around for quite some time – since 1997! – and is maintained by the Data Mining Group.
Many vendors support the PMML standard and the community is still actively contributing to it.
So why should you care about PMML? Well, it is one of the ways to productize the results of your Data Science. Some consider this to be the hardest part. A PMML file enables sharing of predictive analytics models between different applications, making it possible to, for example, build a model in one system, move it to another system to test its performance against a test data set, and then move it to APM Studio for inclusion in your application or solution.
There are other ways to productize your results (such as Python Pickles, POJO and MOJOs) but these require programming knowledge and specific interfacing, whereas PMML support is provided by default in APM Studio.
A ML model trained and validated in KNIME to classify the oil quality based on available data:
The PMML loaded in APM Studio and connected to streaming data feeds:
Curated from UReason
Get your Training Brochure here!