Loading and executing a PMML file

The Predictive Model Markup Language (PMML) is an XML-based file format developed by the Data Mining Group to provide a way for applications to describe and exchange models produced by data mining and machine learning algorithms.

Yacs provides an elementary python node called PyLoadPMML that generates an object of type pyobj from a model read in a PMML file. This pyobj is a PyFunction that can be executed in a python node created by the user.

Authorized PMML model types

Node PyLoadPMML uses the swig/python interface to library libpmmlLib.so (Linux) or pmmllib.dll (Windows). This library handles :

• Neuronal Network models and

• Linear Regression models.

Description of PyLoadPMML

• Input ports

Input port name	YACS type	Comment
filename	string	Name of the PMML file, including its path if the file is not in the current directory
modelname	string	Name of the model to load
pmmltype	string	Type of the model. Value is one of kLR (linear regression) or kANN (neural network)

• Output ports

Output port name	YACS type	Comment
pyFunc	pyobj	PyFunction representing the model This function takes a vector of doubles as input parameter and returns a value of type double

Example of model execution

Create a YACS schema that uses node PyLoadPMML and add a python node that will execute the the pyfunction created by PyLoadPMML. The YACS schema with the execution python node code is shown below :

The characteristics of the execution node are the following:

• Input ports

Input port name	YACS type	Comment
myFunc	pyobj	Linked to the PyFunction generated by PyLoadPMML
params	dblevec	Vector of doubles, input of the PyFunction

• Output ports

Output port name	YACS type	Comment
o5	double	Result of the model execution