x_train, x_test, y_train, y_test = train_test_split(x, y, train_size=0.85, random_state=1)model = PMMLPipeline([('LogisticModer', LogisticRegression())])model.fit(x_train, y_train)y_hat = model.predict(x_test)loss = y_hat == y_testaccuracy = np.mean(loss)print(accuracy)sklearn2pmml(model, '.\LogisticRegression.pmml', with_repr=True)

from sklearn2pmml import sklearn2pmml
from sklearn2pmml.pipeline import PMMLPipeline

 1         <dependency>
 2             <groupId>org.jpmml</groupId>
 3             <artifactId>pmml-evaluator</artifactId>
 4             <version>1.4.1</version>
 5         </dependency>
 6
 7
 8         <dependency>
 9             <groupId>org.jpmml</groupId>
10             <artifactId>pmml-evaluator-extension</artifactId>
11             <version>1.4.1</version>
12         </dependency>
13         <dependency>

public static PMML getPMMLModel(InputStream inputStream) {PMML pmml = new PMML();try {pmml = org.jpmml.model.PMMLUtil.unmarshal(inputStream);} catch (SAXException e1) {e1.printStackTrace();} catch (JAXBException e2) {e2.printStackTrace();} finally {try {inputStream.close();} catch (IOException e) {e.printStackTrace();}return pmml;}}public static Evaluator loadPmmlAndgetEvaluator(MachineLearnType machineLearnType) {String modefile = getJpmmlModelPath(machineLearnType);  //获取模型的pmml文件路径
InputStream inputStream = readPmmlFile(modefile);  //根据文件路径返回输入流
PMML pmml = getPMMLModel(inputStream);  //根据输入流返回PMML
ModelEvaluatorFactory modelEvaluatorFactory = ModelEvaluatorFactory.newInstance();  //获取 ModelEvaluatorFactory
Evaluator evaluator = modelEvaluatorFactory.newModelEvaluator(pmml);  // 根据 PMML 模型返回 Evaluator 对象
pmml = null;return evaluator;}public static Map<String, Object> modelPrediction(Evaluator evaluator, Map<String, Object> paramData) {if (evaluator == null || paramData == null) {System.out.println("--------------传入对象 evaluator 或 dataMap 为空, 无法进行预测----------------");return null;}List<InputField> inputFields = evaluator.getInputFields();   //获取模型的输入域Map<FieldName, FieldValue> arguments = new LinkedHashMap<>();for (InputField inputField : inputFields) {            //将参数通过模型对应的名称进行添加FieldName inputFieldName = inputField.getName();   //获取模型中的参数名Object paramValue = paramData.get(inputFieldName.getValue());   //获取模型参数名对应的参数值FieldValue fieldValue = inputField.prepare(paramValue);   //将参数值填入模型中的参数中arguments.put(inputFieldName, fieldValue);          //存放在map列表中
        }Map<FieldName, ?> results = evaluator.evaluate(arguments);List<TargetField> targetFields = evaluator.getTargetFields();Map<String, Object> resultMap = new HashMap<>();for(TargetField targetField : targetFields) {FieldName targetFieldName = targetField.getName();Object targetFieldValue = results.get(targetFieldName);if (targetFieldValue instanceof Computable) {Computable computable = (Computable) targetFieldValue;resultMap.put(targetFieldName.getValue(), computable.getResult());}else {resultMap.put(targetFieldName.getValue(), targetFieldValue);}}return resultMap;}

Evaluator evaluator = JPmmlModelUtil.loadPmmlAndgetEvaluator(MachineLearnType.LOGISTIC_REGRESSION);
Map<String , Object> results = JPmmlModelUtil.modelPrediction(evaluator, paramMap);
int result =(int)((double)results.get("y"));

　　/*** 使用pmml方式对输入的参数进行线性回归预测*/@PostMapping("/logispmml")public ServerResponse<String> IrisLogosPmmlPredict(@RequestParam @Valid double x1,@RequestParam @Valid double x2,@RequestParam @Valid double x3,@RequestParam @Valid double x4) {logger.info("x1: " + x1 + " x2: " + x2 + " x3:" + x3 + "x4:" + x4);Map<String, Object> paramMap = new HashMap<>();paramMap.put("x1", x1);paramMap.put("x2", x2);paramMap.put("x3", x3);paramMap.put("x4", x4);String result = logisticRegressionService.pridictlogisticpmml(paramMap);return createBySuccess(result);}

int result =(int)((double)results.get("y"));
String irisName = new String();
if(result == 0){irisName = "Iris-setosa";
}
if(result == 1){irisName = "Iris-versicolor";
}
if(result == 2){irisName = "Iris-virginica";
}return irisName;
}