JPMML-evaluator简单示例代码和说明

概要

书接上文，在简单的跑通鸢尾花示例代码之后要拿自己的模型下手了，在经历一番艰难调整之后总算跑通了第一个无监督学习模型。本例采用sklearn2pmml生成的pmml模型，模型由pmml.pipeline流水线构成，包含PCA方法和KMEANS方法，通过java的JPMML库进行调用，预测新的成员属于哪个群体。

代码

import org.dmg.pmml.FieldName;
import org.jpmml.evaluator.*;
import java.io.*;
import java.util.*;


public class TestPmml {
    public static void main(String args[]) throws Exception {
        String fp = "edu.pmml";
        TestPmml obj = new TestPmml();
        Evaluator model = new LoadingModelEvaluatorBuilder()
                .load(new File("edu.pmml"))
                .build();        #模型初始化
        Map<FieldName, Object> arguments = new HashMap<>(); #格式化输入
        arguments.put(FieldName.create("Aggression"), 35.6058);
        arguments.put(FieldName.create("Stress"), 24.193);
        arguments.put(FieldName.create("Tension"), 47.7854);
        arguments.put(FieldName.create("Suspect"), 36.18);
        arguments.put(FieldName.create("Balance"), 77.5673);
        arguments.put(FieldName.create("Charm"), 78.1811);
        arguments.put(FieldName.create("Energy"), 25.8678);
        arguments.put(FieldName.create("Self-Regulation"), 77.715);
        arguments.put(FieldName.create("Inhibition"), 25.4405);
        arguments.put(FieldName.create("Neuroticism"), 34.7914);
        arguments.put(FieldName.create("Extraversion"), 0.619428);
        arguments.put(FieldName.create("Stability"), 0.509011);
        arguments.put(FieldName.create("Depression"), 20.4566);
        arguments.put(FieldName.create("Happiness"), 37.0281);
        arguments.put(FieldName.create("FH_M"), 1.98672);
        arguments.put(FieldName.create("FH_S"), 0.630348);
        #模型评估
        Map<FieldName, ?> results = model.evaluate(arguments);

        List<TargetField> targetFields = model.getTargetFields();
        #获取结果
        for (TargetField targetField : targetFields) {
            Object targetValue = results.get(targetField.getName());
            System.out.println("target=" + EvaluatorUtil.decode(targetValue));

            // 回归和分类模型的结果区域
            if (targetValue instanceof HasReport) {
                HasReport hasReport = (HasReport) targetValue;

                Report report = hasReport.getReport();
                if (report != null) {
                    System.out.println("target=" + ReportUtil.format(report));
                }
            } // End if

            // 概率分类模型的结果区域
            if (targetValue instanceof HasProbability hasProbability) {

                Set<Object> targetCategories = hasProbability.getCategories();
                for (Object targetCategory : targetCategories) {
                    Double probability = hasProbability.getProbability(targetCategory);
                    System.out.println("probability(" + targetCategory + ")=" + probability);

                    Report probabilityReport = hasProbability.getProbabilityReport(targetCategory);
                    if (probabilityReport != null) {
                        System.out.println("probability(" + targetCategory + ")=" + ReportUtil.format(probabilityReport));
                    }
                }
            }
        }
    }
}

本例重点在于格式化的输入和输出，JPMML规定了一个标准化的输入方法，即Map<FieldName, Object> arguments。标准化的结果获取，即evaluate方法和getTargetFields方法。