hadoop集群运行MR程序、mahout程序

本教程在配置完hadoop，可以正常运行的前提下进行

1. 启动集群

# 启动hdfs
sbin/start-dfs.sh

# 启动yarn
sbin/start-yarn.sh

使用jps命令，看到如下图所示，启动成功。

---

2. 上传程序资源到hdfs

1. 第一步：把文件上传到服务器。
2. 第二步：把文件上传到hdfs集群。

bin/hadoop dfs -put ../train_data /

看到如下图所示，上传成功。

---

3. 修改程序文件路径

把程序读取文件的路径，修改为从参数中获取，本程序需要修改为如下代码：

package com.demo;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;

import java.io.IOException;
import java.util.*;

/**
 * @author affable
 * @description 处理数据
 * @date 2020/5/13 10:30
 */
public class DataAnalysis {

    /**
     * 解析文件，取出userId
     */
    static class ParseTxtMapper extends Mapper<LongWritable, Text, Text, Text> {

        Text k = new Text();
        Text v = new Text();

        @Override
        protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
            // 切分获取每个字段值
            String[] fields = value.toString().split(",");

            // k -> userId
            // v -> workId,action
            k.set(fields[0]);
            v.set(fields[1] + "," + fields[2]);
            context.write(k, v);
        }
    }

    /**
     * 按每个userId进行reducer
     */
    static class ScoreReducer extends Reducer<Text, Text, Text, Text> {

        // 此userId对于所有浏览职位的action
        Map<String, Integer> workActions = new HashMap<>();
        // 此userId对于所有已投递职位的action
        Map<String, Integer> deliveryWorks = new HashMap<>();

        // 最大值
        double maxWorkAction = 0.0D;
        double maxDeliveryWorkAction = 0.0D;

        // 这次的userId
        String userId;

        Text k = new Text();
        Text v = new Text();

        @Override
        protected void reduce(Text key, Iterable<Text> values, Context context) throws IOException, InterruptedException {
            userId = key.toString();
            for (Text value : values) {
                String[] fields = value.toString().split(",");
                if ("2".equals(fields[1])) {
                    // 如果投递了该职位
                    // 如果map中没有，则放入1，如果有，则在此基础加1
                    deliveryWorks.merge(fields[0], 1, Integer::sum);
                    Integer count = deliveryWorks.get(fields[0]);
                    maxDeliveryWorkAction = count > maxDeliveryWorkAction ? count : maxDeliveryWorkAction;
                } else {
                    // 只是浏览了职位
                    workActions.merge(fields[0], 1, Integer::sum);
                    Integer count = workActions.get(fields[0]);
                    maxWorkAction = count > maxWorkAction ? count : maxWorkAction;
                }
            }

            // 从浏览职位中去除已投递的
            deliveryWorks.forEach((k, v) -> workActions.remove(k));

            // 计算分数
            // 规则:
            //     浏览量/最大浏览量*4  (0, 4]
            //     投递量/最大投递量+4  (4, 5]
            for (Map.Entry<String, Integer> entry : workActions.entrySet()) {
                String workId = entry.getKey();
                Integer count = entry.getValue();
                k.set(userId + "\t" + workId);
                v.set(String.format("%.2f", count / maxWorkAction * 4));
                context.write(k, v);
            }

            for (Map.Entry<String, Integer> entry : deliveryWorks.entrySet()) {
                String workId = entry.getKey();
                Integer count = entry.getValue();
                k.set(userId + "\t" + workId);
                v.set(String.format("%.2f", 4.0 + count / maxDeliveryWorkAction));
                context.write(k, v);
            }

            // 清空数据
            workActions.clear();
            deliveryWorks.clear();
            maxWorkAction = 0.0D;
            maxDeliveryWorkAction = 0.0D;
        }

    }

    public static void main(String[] args) throws Exception {

        String inputPath = args[0];
        String outputPath = args[1];

        Configuration conf = new Configuration();

        Job job = Job.getInstance(conf, "cal_user_score");
        job.setJarByClass(DataAnalysis.class);
        job.setMapperClass(ParseTxtMapper.class);
        job.setReducerClass(ScoreReducer.class);
        job.setOutputKeyClass(Text.class);
        job.setOutputValueClass(Text.class);

        job.setNumReduceTasks(1);

        // 设置输入输出路径
        FileInputFormat.setInputPaths(job, new Path(inputPath));
        FileOutputFormat.setOutputPath(job, new Path(outputPath));


        System.exit(job.waitForCompletion(true) ? 0 : 1);
    }

}

package com.demo;

import org.apache.commons.csv.CSVParser;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.mahout.cf.taste.common.TasteException;
import org.apache.mahout.cf.taste.impl.common.FastByIDMap;
import org.apache.mahout.cf.taste.impl.common.LongPrimitiveIterator;
import org.apache.mahout.cf.taste.impl.model.GenericDataModel;
import org.apache.mahout.cf.taste.impl.model.GenericPreference;
import org.apache.mahout.cf.taste.impl.model.GenericUserPreferenceArray;
import org.apache.mahout.cf.taste.impl.model.MemoryIDMigrator;
import org.apache.mahout.cf.taste.impl.recommender.svd.ALSWRFactorizer;
import org.apache.mahout.cf.taste.impl.recommender.svd.SVDRecommender;
import org.apache.mahout.cf.taste.model.DataModel;
import org.apache.mahout.cf.taste.model.Preference;
import org.apache.mahout.cf.taste.model.PreferenceArray;
import org.apache.mahout.cf.taste.recommender.RecommendedItem;
import org.apache.mahout.cf.taste.recommender.Recommender;

import java.io.*;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;

/**
 * @author affable
 * @description 职位推荐
 * @date 2020/5/11 10:02
 */
public class WorkRecommend {

    private static final int SIZE = 20;

    /**
     * 获取文件系统
     */
    private static FileSystem getFiledSystem() throws IOException {
        Configuration configuration = new Configuration();
        return FileSystem.get(configuration);
    }

    /**
     * 读取hdfs文件
     * @param filePath 文件路径
     * @return 返回读取文件流
     */
    private static BufferedReader readHDFSFile(String filePath) throws FileNotFoundException {
        FSDataInputStream fsDataInputStream = null;
        BufferedReader reader = null;
        try {
            Path path = new Path(filePath);
            fsDataInputStream = getFiledSystem().open(path);
            reader = new BufferedReader(new InputStreamReader(fsDataInputStream));
        } catch (IOException e) {
            e.printStackTrace();
        }
        return reader;
    }


    public static void main(String[] args) throws TasteException, IOException {
        // *******************************处理开始******************************************
        // 使用推荐模型之前，对数据的userId和workId映射成long类型
        // 防止模型把userId和workId转为long，出现数据异常
        // 数据存储路径
        String dataPath = args[0];
        Map<Long,List<Preference>> preferecesOfUsers = new HashMap<>(16);
        // 读取原始数据并处理
        BufferedReader reader = readHDFSFile(dataPath);
        CSVParser parser = new CSVParser(reader, '\t');
        String[] line;
        MemoryIDMigrator userIdMigrator = new MemoryIDMigrator();
        MemoryIDMigrator workIdMigrator = new MemoryIDMigrator();
        while((line = parser.getLine()) != null) {
            // string 转 long
            long userIdLong = userIdMigrator.toLongID(line[0]);
            long workIdLong = workIdMigrator.toLongID(line[1]);
            userIdMigrator.storeMapping(userIdLong, line[0]);
            workIdMigrator.storeMapping(workIdLong, line[1]);
            List<Preference> userPrefList;
            if((userPrefList = preferecesOfUsers.get(userIdLong)) == null) {
                userPrefList = new ArrayList<>();
                preferecesOfUsers.put(userIdLong, userPrefList);
            }
            userPrefList.add(new GenericPreference(userIdLong, workIdLong, Float.parseFloat(line[2])));
        }
        FastByIDMap<PreferenceArray> preferecesOfUsersFastMap = new FastByIDMap<>();
        for(Map.Entry<Long, List<Preference>> entry : preferecesOfUsers.entrySet()) {
            preferecesOfUsersFastMap.put(entry.getKey(), new GenericUserPreferenceArray(entry.getValue()));
        }
        // ***********************************处理完成**************************************


        // 读取数据
        DataModel dataModel = new GenericDataModel(preferecesOfUsersFastMap);
        // 使用als求损失函数
        ALSWRFactorizer factorizer = new ALSWRFactorizer(dataModel, 5, 0.2, 200);
        // 使用SVD算法进行推荐
        Recommender recommender = new SVDRecommender(dataModel, factorizer);

        // 推荐测试
        LongPrimitiveIterator userIdIterator = dataModel.getUserIDs();
        int i = 0;
        while (userIdIterator.hasNext()) {
            long userIdLong = userIdIterator.nextLong();
            String userId = userIdMigrator.toStringID(userIdLong);
            List<RecommendedItem> recommendedItems = recommender.recommend(userIdLong, SIZE);
            for (RecommendedItem item : recommendedItems) {
                // 写出到mysql
                // MysqlUtils.insert(userId, workIdMigrator.toStringID(item.getItemID()), item.getValue());
                System.out.println(String.format("userId: %s, itemId: %s, score: %.2f",
                        userId, workIdMigrator.toStringID(item.getItemID()), item.getValue()));
                i++;
            }
        }
        System.out.println(i);

        // 释放资源
        // MysqlUtils.release();
        reader.close();

    }

}

---

4. 安装mahout

1. 下载安装包。
2. 上传到服务器，并解压。
3. 修改环境变量，在/etc/profile中做如下修改，并重新加载profile文件。

source /etc/profile

5. 修改mahout配置（或者添加HADOOP_CONF_DIR环境变量也可以）

vim bin/mahout

6. 在命令行输入 mahout，测试是否可以正常运行。

---

5. 提交程序到集群

1. 本地使用以下命令对程序打包，并上传到服务器。
   最好把后缀为-with-dependencies.jar的包上传到服务器，防止hadoop集群没有对应的依赖包

# maven程序打包
mvn clean package

2. 运行数据分析处理程序。

# 最后两个参数为：待处理数据路径，处理后结果的保存路径
bin/hadoop jar ../work-recommend-1.0.0-Release-jar-with-dependencies.jar com.demo.DataAnalysis /train_data/user_action.csv /work_out/user_score

如下图所示，运行成功。


3. 运行预测程序。

mahout hadoop jar ../work-recommend-1.0.0-Release-jar-with-dependencies.jar com.demo.WorkRecommend hdfs://localhost:9000/work_out/user_score/part-r-00000

执行结果如下