0

jpmml

Java PMML API (legacy codebase)

Github 资源库
主要指标
  • 所有者: jpmml/jpmml
  • 平台:
  • 许可证: Other
  • 分类:
  • 主题:
  • 喜欢:
    0
      比较:

Github星跟踪图

Java API for producing and scoring models in Predictive Model Markup Language (PMML).

IMPORTANT

This is a legacy codebase.

Starting from March 2014, this project has been superseded by [JPMML-Model] (https://github.com/jpmml/jpmml-model) and [JPMML-Evaluator] (https://github.com/jpmml/jpmml-evaluator) projects.

Features

Class model

  • Full support for PMML 3.0, 3.1, 3.2, 4.0 and 4.1 schemas:
    • Class hierarchy.
    • Schema version annotations.
  • Fluent API:
    • Value constructors.
  • SAX Locator information
  • [Visitor pattern] (http://en.wikipedia.org/wiki/Visitor_pattern):
    • Validation agents.
    • Optimization and transformation agents.

Evaluation engine

Installation

JPMML library JAR files (together with accompanying Java source and Javadocs JAR files) are released via [Maven Central Repository] (http://repo1.maven.org/maven2/org/jpmml/). Please join the [JPMML mailing list] (https://groups.google.com/forum/#!forum/jpmml) for release announcements.

The current version is 1.0.22 (17 February, 2014).

Class model

<!-- Class model classes -->
<dependency>
	<groupId>org.jpmml</groupId>
	<artifactId>pmml-model</artifactId>
	<version>${jpmml.version}</version>
</dependency>
<!-- Class model annotations -->
<dependency>
	<groupId>org.jpmml</groupId>
	<artifactId>pmml-schema</artifactId>
	<version>${jpmml.version}</version>
</dependency>

Evaluation engine

<dependency>
	<groupId>org.jpmml</groupId>
	<artifactId>pmml-evaluator</artifactId>
	<version>${jpmml.version}</version>
</dependency>

Usage

Class model

The class model consists of two types of classes. There is a small number of manually crafted classes that are used for structuring the class hierarchy. They are permanently stored in the Java sources directory /pmml-model/src/main/java. Additionally, there is a much greater number of automatically generated classes that represent actual PMML elements. They can be found in the generated Java sources directory /pmml-model/target/generated-sources/xjc after a successful build operation.

All class model classes descend from class org.dmg.pmml.PMMLObject. Additional class hierarchy levels, if any, represent common behaviour and/or features. For example, all model classes descend from class org.dmg.pmml.Model.

There is not much documentation accompanying class model classes. The application developer should consult with the [PMML specification] (http://www.dmg.org/v4-1/GeneralStructure.html) about individual PMML elements and attributes.

Example applications

Evaluation engine

A model evaluator class can be instantiated directly when the contents of the PMML document is known:

PMML pmml = ...;

ModelEvaluator<TreeModel> modelEvaluator = new TreeModelEvaluator(pmml);

Otherwise, a PMML manager class should be instantiated first, which will inspect the contents of the PMML document and instantiate the right model evaluator class later:

PMML pmml = ...;

PMMLManager pmmlManager = new PMMLManager(pmml);
 
ModelEvaluator<?> modelEvaluator = (ModelEvaluator<?>)pmmlManager.getModelManager(null, ModelEvaluatorFactory.getInstance());

Model evaluator classes follow functional programming principles. Model evaluator instances are cheap enough to be created and discarded as needed (ie. not worth the pooling effort).

It is advisable for application code to work against the org.jpmml.evaluator.Evaluator interface:

Evaluator evaluator = (Evaluator)modelEvaluator;

An evaluator instance can be queried for the definition of active (ie. independent), predicted (ie. primary dependent) and output (ie. secondary dependent) fields:

List<FieldName> activeFields = evaluator.getActiveFields();
List<FieldName> predictedFields = evaluator.getPredictedFields();
List<FieldName> outputFields = evaluator.getOutputFields();

The PMML scoring operation must be invoked with valid arguments. Otherwise, the behaviour of the model evaluator class is unspecified.

The preparation of field values:

Map<FieldName, FieldValue> arguments = new LinkedHashMap<FieldName, FieldValue>();

List<FieldName> activeFields = evaluator.getActiveFields();
for(FieldName activeField : activeFields){
	// The raw (ie. user-supplied) value could be any Java primitive value
	Object rawValue = ...;

	// The raw value is passed through: 1) outlier treatment, 2) missing value treatment, 3) invalid value treatment and 4) type conversion
	FieldValue activeValue = evaluator.prepare(activeField, rawValue);

	arguments.put(activeField, activeValue);
}

The scoring:

Map<FieldName, ?> results = evaluator.evaluate(arguments);

Typically, a model has exactly one predicted field, which is called the target field:

FieldName targetName = evaluator.getTargetField();
Object targetValue = results.get(targetName);

The target value is either a Java primitive value (as a wrapper object) or an instance of org.jpmml.evaluator.Computable:

if(targetValue instanceof Computable){
	Computable computable = (Computable)targetValue;

	Object primitiveValue = computable.getResult();
}

The target value may implement interfaces that descend from interface org.jpmml.evaluator.ResultFeature:

// Test for "entityId" result feature
if(targetValue instanceof HasEntityId){
	HasEntityId hasEntityId = (HasEntityId)targetValue;
	HasEntityRegistry<?> hasEntityRegistry = (HasEntityRegistry<?>)evaluator;
	BiMap<String, ? extends Entity> entities = hasEntityRegistry.getEntityRegistry();
	Entity winner = entities.get(hasEntityId.getEntityId());

	// Test for "probability" result feature
	if(targetValue instanceof HasProbability){
		HasProbability hasProbability = (HasProbability)targetValue;
		Double winnerProbability = hasProbability.getProbability(winner.getId());
	}
}
Example applications

Additional information

Please contact [info@openscoring.io] (mailto:info@openscoring.io)

主要指标

概览
名称与所有者jpmml/jpmml
主编程语言Java
编程语言Java (语言数: 3)
平台
许可证Other
所有者活动
创建于2013-02-27 12:54:09
推送于2015-06-16 09:33:58
最后一次提交2014-03-08 08:43:48
发布数23
最新版本名称1.0.22 (发布于 2014-02-16 23:41:46)
第一版名称1.0.0 (发布于 2013-02-27 14:56:50)
用户参与
星数80
关注者数12
派生数43
提交数488
已启用问题?
问题数0
打开的问题数0
拉请求数0
打开的拉请求数0
关闭的拉请求数5
项目设置
已启用Wiki?
已存档?
是复刻?
已锁定?
是镜像?
是私有?