Toolkit com.ibm.streams.pmml 2.1.0
Specialized toolkits - release 4.3.1.0-prod20190605 > com.ibm.streams.pmml 2.1.0
General Information
Overview
The PMML toolkit provides operators to score input records using PMML models and to interact with the WML repository.
For details on PMML see Predictive Model Markup Language. For details on WML see Watson Machine Learning Service
The toolkit provides the following operators
- PMMLScoring The operator maps tuple attributes to model predictors and emit the predicted values as output. The model can be spcified in several ways.
- WMLModelFeed This operator loads models and new versions of models from the WML repository, for usage in the PMMLScoring operator
The toolkit supports the following features
- Flexible mapping of Streams attributes to model predictors and output fields
- Model meta-data is included in the output of the PMMLScoring and WMLModelFeed operators
- The model to fetch can be specified by either the model name or the model GUID
- The scoring operator can block tuple processing until an initial model is loaded from WML (with a configurable timeout)
Prerequisites and restrictions
- To build and run applications with this toolkit, the Anaconda 4.2.0 (with Python 3.5.2) Python package must be installed.
Additional information
- Version
- 2.1.0
- Required Product Version
- 4.3.0.0