A library that allows serialization of SciKit-Learn estimators into PMML
Clone or download
alex-pirozhenko Merge pull request #23 from YuHuaCheng/master
fix missing values in categorical features & update version
Latest commit 82e4621 Oct 12, 2015

README.md

Build Status Join the chat at https://gitter.im/alex-pirozhenko/sklearn-pmml

sklearn-pmml

A library that allows serialization of SciKit-Learn estimators into PMML

Installation

The easiest way is to use pip:

pip install sklearn-pmml

Supported models

  • DecisionTreeClassifier
  • DecisionTreeRegressor
  • GradientBoostingClassifier
  • RandomForestClassifier

PMML output

Classification

Classifier converters can only operate with categorical outputs, and for each categorical output variable varname the PMML output contains the following outputs:

  • categorical varname for the predicted label of the instance
  • double varname.label for the probability for a given label

Regression

Regression model PMML outputs the numeric response variable named as the output variable