數據挖掘-Mahout-Canopy聚類實踐

1、原理解釋

(1)原始數據集合List按照一定的規則進行排序，初始距離閾值設置為T1、T2，T1>T2。

(2)在List中隨機挑選一個數據向量A，使用一個粗糙距離計算方式計算A與List中其它樣本數據向量之間的距離d。

(3)根據2中的距離d，把d小于T1的樣本數據向量劃到一個canopy中，同時把d小于T2的樣本數據向量從List中移除。

(4)重復2、3，直至List為空

2、下載測試數據

cd /tmp

hadoop dfs -mkdir /input

wget http://archive.ics.uci.edu/ml/databases/synthetic_control/synthetic_control.data

hadoop dfs -copyFromLocal /tmp/synthetic_control.data /input/synthetic_control.data

3、格式轉換（文本→向量）

編輯文件 Text2VectorWritable.jar

package mahout.fansy.utils.transform;

import java.io.IOException;

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.SequenceFileOutputFormat;

import org.apache.hadoop.util.ToolRunner;

import org.apache.mahout.common.AbstractJob;

import org.apache.mahout.math.RandomAccessSparseVector;

import org.apache.mahout.math.Vector;

import org.apache.mahout.math.VectorWritable;

/**

 *  --* transform text data to vectorWritable data

 *   --* @author fansy

 *    --*

 *     --*/

public class Text2VectorWritable extends AbstractJob{

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

       ToolRunner.run(new Configuration(), new Text2VectorWritable(),args);

  }

  @Override

  public int run(String[] arg0) throws Exception {

       addInputOption();

       addOutputOption();

       if (parseArguments(arg0) == null) {

         return -1;

       }

       Path input=getInputPath();

       Path output=getOutputPath();

       Configuration conf=getConf();

       // set job information

       Job job=new Job(conf,"text2vectorWritableCopy with input:"+input.getName());

       job.setOutputFormatClass(SequenceFileOutputFormat.class);

       job.setMapperClass(Text2VectorWritableMapper.class);

       job.setMapOutputKeyClass(LongWritable.class);

       job.setMapOutputValueClass(VectorWritable.class);

       job.setReducerClass(Text2VectorWritableReducer.class);

       job.setOutputKeyClass(LongWritable.class);

       job.setOutputValueClass(VectorWritable.class);

job.setJarByClass(Text2VectorWritable.class);

       FileInputFormat.addInputPath(job, input);

       SequenceFileOutputFormat.setOutputPath(job, output);

       if (!job.waitForCompletion(true)) { // wait for the job is done

            throw new InterruptedException("Canopy Job failed processing " + input);

       }

       return 0;

   }

   /**

    * Mapper main procedure

    * @author fansy

    *

   --*/

   public static class Text2VectorWritableMapper extends Mapper<LongWritable,Text,LongWritable,VectorWritable>{

        public void map(LongWritable key,Text value,Context context) throws IOException,InterruptedException{

            String[] str=value.toString().split("\\s{1,}");

            // split data use one or more blanker

            Vector vector=new RandomAccessSparseVector(str.length);

            for(int i=0;i<str.length;i++){

                 vector.set(i, Double.parseDouble(str[i]));

            }

            VectorWritable va=new VectorWritable(vector);

            context.write(key, va);

       }

   }

   /**

   * Reducer: do nothing but output

   * @author fansy

   *

   --*/

   public static class Text2VectorWritableReducer extends Reducer<LongWritable,VectorWritable,LongWritable,VectorWritable>{

       public void reduce(LongWritable key,Iterable<VectorWritable> values,Context context)throws IOException,InterruptedException{

           for(VectorWritable v:values){

                context.write(key, v);

           }

       }

   }

}

編譯，輸出ClusteringUtils.jar，并拷貝至/home/mahout/mahout_jar

輸出時選擇Export→Runnable Jar File→Extract required libraries into generated JAR

然后執行：

hadoop jar /home/hadoop/mahout/mahout_jar/ClusteringUtils.jar mahout.fansy.utils.transform.Text2VectorWritable -i hdfs:///input/synthetic_control.data -o hdfs:///input/synthetic_control.data.transform

有可能會遇到org/apache/mahout/common/AbstractJob找不到類報錯，這個一般是由于HADOOP_CLASSPATH配置位置不包含mahout的jar的原因。

解決方法1：

拷貝mahout的jar文件到/home/hadoop/lib中去，并確認這個/home/hadoop/lib確實在HADOOP_CLASSPATH中

cp /home/hadoop/mahout/*.jar /home/hadoop/hadoop/lib

解決方法2（推薦）：

在hadoop-env.sh中加入

for f in /home/hadoop/mahout/*.jar; do

  if [ "$HADOOP_CLASSPATH" ]; then

    export HADOOP_CLASSPATH=$HADOOP_CLASSPATH:$f

  else

    export HADOOP_CLASSPATH=$f

  fi

done

記得將hadoop-evn.sh分發到其它節點

重啟hadoop環境

stop-all.sh

start-all.sh

執行轉換：

hadoop jar /home/hadoop/mahout/mahout_jar/ClusteringUtils.jar mahout.fansy.utils.transform.Text2VectorWritable -i hdfs:///input/synthetic_control.data -o hdfs:///input/synthetic_control.data.transform（如果在導出Jar的時候已經指派主類，這個命令會報錯，使用下面的命令）

hadoop jar /home/hadoop/mahout/mahout_jar/ClusteringUtils.jar -o hdfs:///input/synthetic_control.data.transform

輸出完畢的文件已經是面目全非的Vector文件了

hdfs:///input/synthetic_control.data.transform/part-r-00000

4、執行Canopy聚類

mahout canopy --input hdfs:///input/synthetic_control.data.transform/part-r-00000 --output /output/canopy --distanceMeasure org.apache.mahout.common.distance.EuclideanDistanceMeasure --t1 80 --t2 55 --t3 80 --t4 55 --clustering

5、轉換格式（向量→文本）

把4中的結果轉換成為文本

編輯文件ReadClusterWritable.java

package mahout.fansy.utils;  

import java.io.IOException;  

import org.apache.hadoop.conf.Configuration;  

import org.apache.hadoop.io.Text;  

import org.apache.hadoop.mapreduce.Job;  

import org.apache.hadoop.mapreduce.Mapper;  

import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;  

import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;  

import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;  

import org.apache.hadoop.util.ToolRunner;  

import org.apache.mahout.clustering.iterator.ClusterWritable;  

import org.apache.mahout.common.AbstractJob;  

import org.slf4j.Logger;  

import org.slf4j.LoggerFactory;  

/**  

 ??* read cluster centers  

 ??* @author fansy  

 ??*/  

public class ReadClusterWritable extends AbstractJob {  

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

           ToolRunner.run(new Configuration(), new ReadClusterWritable(),args);  

      }  

      @Override  

      public int run(String[] args) throws Exception {  

           addInputOption();  

           addOutputOption();  

           if (parseArguments(args) == null) {  

                return -1;  

             }  

           Job job=new Job(getConf(),getInputPath().toString());  

           job.setInputFormatClass(SequenceFileInputFormat.class);  

           job.setMapperClass(RM.class);  

           job.setMapOutputKeyClass(Text.class);  

           job.setMapOutputValueClass(Text.class);  

           job.setNumReduceTasks(0);  

           job.setJarByClass(ReadClusterWritable.class);  

           FileInputFormat.addInputPath(job, getInputPath());  

           FileOutputFormat.setOutputPath(job, getOutputPath());  

           if (!job.waitForCompletion(true)) {  

              throw new InterruptedException("Canopy Job failed processing " + getInputPath());  

        }  

      return 0;  

    }  

    public static class RM extends Mapper<Text,ClusterWritable ,Text,Text>{  

         private Logger log=LoggerFactory.getLogger(RM.class);  

       public void map(Text key,ClusterWritable value,Context context) throws  

IOException,InterruptedException{  

              String str=value.getValue().getCenter().asFormatString();  

         //   System.out.println("center****************:"+str);  

           log.info("center*****************************:"+str); // set log information  

              context.write(key, new Text(str));  

         }  

    }  

}

打包到ClusteringUtils.jar，上傳到/home/hadoop/mahout/mahout_jar

如果需要清除eclipse中Launch Configuration中的信息，需要進入工程所在文件夾下的/.metadata/.plugins/org.eclipse.debug.core/.launches

然后刪除里面的文件

運行

hadoop jar ClusteringUtils.jar mahout.fansy.utils.ReadClusterWritable -i /output/canopy/clusters-0-final/part-r-00000 -o /output/canopy-output（如果不成功就運行下面的命令）

hadoop jar ClusteringUtils.jar -i /output/canopy/clusters-0-final/part-r-00000 -o /output/canopy-output

這時候/output/canopy-output/part-m-00000里面放置的就是聚類的結果文件