基于Kubernetes、Docker的机器学习微服务系统设计系列——(六)特征选择微服务

最新推荐文章于 2022-10-18 10:30:12 发布

原创

最新推荐文章于 2022-10-18 10:30:12 发布 · 594 阅读

0 ·

CC 4.0 BY-SA版权

文章标签：

#机器学习 #特征选择 #JSON #文本分类 #微服务

本文介绍了如何设计一个基于Kubernetes和Docker的机器学习微服务系统，专注于特征选择部分。特征选择微服务涵盖了Document Frequency、Information Gain、Chi-Square Test、Mutual Information和Matrix Projection等算法。文中详细展示了特征选择类图、部分实现代码，以及服务的请求和响应JSON格式。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

特征选择微服务主要实现如下特征选择算法：Document Frequency(DF)、Information Gain(IG)、(χ²)Chi-Square Test(CHI)、Mutual Information(MI)、Matrix Projection(MP)。

特征选择类图

特征选择类图如图所示：

特征选择微服务类图

部分实现代码

特征选择Action类

package com.robin.feature.action;

import com.robin.feature.corpus.CorpusManager;
import com.robin.feature.AbstractFeature;
import com.robin.feature.FeatureFactory;
import com.robin.feature.FeatureFactory.FeatureMethod;
import com.robin.loader.MircoServiceAction;
import com.robin.log.RobinLogger;
import java.util.HashSet;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.logging.Level;
import java.util.logging.Logger;
import org.codehaus.jettison.json.JSONArray;
import org.codehaus.jettison.json.JSONException;
import org.codehaus.jettison.json.JSONObject;

/**
 * <DT><B>描述：</B></DT>
 * <DD>特征选择Action类</DD>
 *
 * 适配Jersey服务器资源调用
 *
 * @version Version1.0
 * @author Robin
 * @version <I>　Date：2018-04-01</I>
 * @author  <I>　E-mail:xsd-jj@163.com</I>
 */
public class FeatureSelectAction implements MircoServiceAction {
   
   

    private static final Logger LOGGER = RobinLogger.getLogger();

    /**
     * Action状态码
     */
    public enum StatusCode {
   
   
        OK,
        JSON_ERR,
        KIND_ERR,
        VERSION_ERR,
        TRAIN_SCALE_ERR,
        METHOD_ERR,
        TEXTS_NULL,
    }

    /**
     * Action状态内部类
     */
    private class ActionStatus {
   
   

        StatusCode statusCode;
        String msg;

    }

    /**
     * 获取返回错误状态JSONObject
     *
     * @param actionStatus
     * @return JSONObject
     */
    private JSONObject getErrorJson(ActionStatus actionStatus) {
   
   
        JSONObject errJson = new JSONObject();
        try {
   
   
            errJson.put("status", actionStatus.statusCode.toString());
            errJson.put("msg", actionStatus.msg);
        } catch (JSONException ex) {
   
   
            LOGGER.log(Level.SEVERE, ex.getMessage());
        }
        return errJson;
    }

    /**
     * 检查JSON输入对象具体项
     *
     * @param jsonObj
     * @param key
     * @param valueSet
     * @param errStatusCode
     * @return ActionStatus
     */
    private ActionStatus checkJSONObjectTerm(JSONObject jsonObj,
            String key,
            HashSet<String> valueSet,
            StatusCode errStatusCode) {
   
   
        ActionStatus actionStatus = new ActionStatus();

        try {
   
   
            if (!jsonObj.isNull(key)) {
   
   
                String value = jsonObj.getString(key);
                if (!valueSet.contains(value)) {
   
   
                    actionStatus.msg = "The value [" + value + "] of " + key + " is error.";
                    actionStatus.statusCode = errStatusCode;
                    return actionStatus;
                }
            } else {
   
   
                actionStatus.msg = "The input parameter is missing " + key + ".";
                actionStatus.statusCode = errStatusCode;
                return actionStatus;
            }

        } catch (JSONException ex) {
   
   
            LOGGER.log(Level.SEVERE, ex.getMessage());
        }

        actionStatus.statusCode = StatusCode.OK;
        return actionStatus;
    }

    /**
     * 检查JSON输入对象
     *
     * @param jsonObj
     * @return ActionStatus
     */
    private ActionStatus checkInputJSONObject(JSONObject jsonObj) {
   
   
        ActionStatus actionStatus = new ActionStatus();
        ActionStatus retActionStatus;

        HashSet<String> valueSet = new HashSet();

        valueSet.add("feature");
        retActionStatus = checkJSONObjectTerm(jsonObj, "kind", valueSet, StatusCode.KIND_ERR);
        if (!retActionStatus.statusCode.equals(StatusCode.OK)) {
   
   
            return retActionStatus;
        }

        valueSet.clear();
        valueSet.add("v1");
        retActionStatus = checkJSONObjectTerm(jsonObj, "version", valueSet, StatusCode.VERSION_ERR);
        if (!retActionStatus.statusCode.equals(StatusCode.OK)) {
   
   
            return retActionStatus;
        }

        try {
   
   
            double trainScale = jsonObj.getJSONObject("metadata").getJSONObject("feature").getDouble("trainScale");
            if ((trainScale >= 1.0) || (trainScale <= 0)) {
   
   
                actionStatus.statusCode = StatusCode.TRAIN_SCALE_ERR;
                actionStatus.msg = "The input train_scale [" + trai