SpringBoot集成Weka机器学习实战

环境准备与依赖配置

在项目的pom.xml文件中添加Weka的Maven依赖：

<dependency>
    <groupId>nz.ac.waikato.cms.weka</groupId>
    <artifactId>weka-stable</artifactId>
    <version>3.8.6</version>
</dependency>

确保SpringBoot基础依赖已配置，推荐使用SpringBoot 2.7.x及以上版本。创建application.yml文件配置服务端口等基础参数：

server:
  port: 8080
spring:
  application:
    name: weka-microservice

模型加载与初始化

在resources目录下存放预训练的Weka模型文件（.model格式）。通过静态代码块初始化模型：

public class WekaService {
    private static Classifier classifier;
    
    static {
        try {
            classifier = (Classifier) SerializationHelper.read(
                Objects.requireNonNull(
                    WekaService.class.getClassLoader()
                        .getResourceAsStream("iris-j48.model")
                )
            );
        } catch (Exception e) {
            throw new RuntimeException("Model loading failed", e);
        }
    }
}

数据预处理模块

创建DTO接收前端传入的预测特征数据：

@Data
public class PredictionRequest {
    private double sepalLength;
    private double sepalWidth;
    private double petalLength;
    private double petalWidth;
}

构建ARFF格式数据转换工具类：

public class ArffBuilder {
    public static Instances createInstance(PredictionRequest request) {
        ArrayList<Attribute> attributes = new ArrayList<>();
        attributes.add(new Attribute("sepal_length"));
        attributes.add(new Attribute("sepal_width"));
        attributes.add(new Attribute("petal_length"));
        attributes.add(new Attribute("petal_width"));
        
        Instances dataset = new Instances("prediction_data", attributes, 0);
        dataset.setClassIndex(attributes.size() - 1);
        
        double[] values = new double[]{
            request.getSepalLength(),
            request.getSepalWidth(),
            request.getPetalLength(),
            request.getPetalWidth()
        };
        dataset.add(new DenseInstance(1.0, values));
        
        return dataset;
    }
}

REST接口实现

创建SpringBoot控制器暴露预测接口：

@RestController
@RequestMapping("/api/predict")
public class PredictionController {
    
    @PostMapping
    public String predict(@RequestBody PredictionRequest request) {
        try {
            Instances instance = ArffBuilder.createInstance(request);
            double result = WekaService.classifier.classifyInstance(instance.firstInstance());
            return instance.classAttribute().value((int) result);
        } catch (Exception e) {
            return "Prediction error: " + e.getMessage();
        }
    }
}

性能优化方案

启用缓存提高重复请求响应速度：

@Cacheable(value = "predictions", key = "#request.toString()")
public String cachedPredict(PredictionRequest request) {
    // 原有预测逻辑
}

配置线程池处理并发请求：

@Configuration
@EnableAsync
public class AsyncConfig {
    @Bean(name = "wekaThreadPool")
    public Executor threadPoolTaskExecutor() {
        ThreadPoolTaskExecutor executor = new ThreadPoolTaskExecutor();
        executor.setCorePoolSize(4);
        executor.setMaxPoolSize(8);
        executor.setQueueCapacity(100);
        executor.setThreadNamePrefix("weka-predict-");
        executor.initialize();
        return executor;
    }
}

异常处理机制

全局异常处理器捕获Weka相关异常：

@ControllerAdvice
public class WekaExceptionHandler {
    
    @ExceptionHandler(WekaException.class)
    @ResponseStatus(HttpStatus.BAD_REQUEST)
    public @ResponseBody String handleWekaException(WekaException ex) {
        return "Weka processing error: " + ex.getMessage();
    }
}

模型热更新方案

实现模型动态加载接口：

@PostMapping("/model/update")
public String updateModel(@RequestParam MultipartFile file) {
    try {
        File tempFile = File.createTempFile("model", ".tmp");
        file.transferTo(tempFile);
        WekaService.reloadModel(tempFile);
        return "Model updated successfully";
    } catch (Exception e) {
        return "Update failed: " + e.getMessage();
    }
}

在WekaService中添加模型重载方法：

public static synchronized void reloadModel(File modelFile) {
    try {
        classifier = (Classifier) SerializationHelper.read(modelFile.getPath());
    } catch (Exception e) {
        throw new WekaException("Model reloading failed", e);
    }
}

微服务监控

集成Prometheus监控指标：

@Bean
public MeterRegistryCustomizer<PrometheusMeterRegistry> metricsCommonTags() {
    return registry -> registry.config().commonTags(
        "application", "weka-microservice"
    );
}

自定义预测指标收集：

@Aspect
@Component
public class PredictionMetricsAspect {
    
    private final Counter predictionCounter = Metrics.counter("weka.predictions.count");
    
    @AfterReturning(pointcut = "execution(* com.example.controller.*.predict(..))", returning = "result")
    public void afterPrediction(Object result) {
        predictionCounter.increment();
    }
}