数据挖掘-Mahout-Canopy聚类实践

1、原理解释

(1)原始数据集合List按照一定的规则进行排序，初始距离阈值设置为T1、T2，T1>T2。

(2)在List中随机挑选一个数据向量A，使用一个粗糙距离计算方式计算A与List中其它样本数据向量之间的距离d。

(3)根据2中的距离d，把d小于T1的样本数据向量划到一个canopy中，同时把d小于T2的样本数据向量从List中移除。

(4)重复2、3，直至List为空

2、下载测试数据

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、格式转换（文本→向量）

编辑文件 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);
}
}
}
}

编译，输出ClusteringUtils.jar，并拷贝至/home/mahout/mahout_jar
输出时选择Export→Runnable Jar File→Extract required libraries into generated JAR

然后执行：
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

有可能会遇到org/apache/mahout/common/AbstractJob找不到类报错，这个一般是由于HADOOP_CLASSPATH配置位置不包含mahout的jar的原因。

解决方法1：
拷贝mahout的jar文件到/home/hadoop/lib中去，并确认这个/home/hadoop/lib确实在HADOOP_CLASSPATH中

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

解决方法2（推荐）：
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

记得将hadoop-evn.sh分发到其它节点

重启hadoop环境
stop-all.sh
start-all.sh

执行转换：

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（如果在导出Jar的时候已经指派主类，这个命令会报错，使用下面的命令）
hadoop jar /home/hadoop/mahout/mahout_jar/ClusteringUtils.jar -o hdfs:///input/synthetic_control.data.transform

输出完毕的文件已经是面目全非的Vector文件了
hdfs:///input/synthetic_control.data.transform/part-r-00000

4、执行Canopy聚类
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、转换格式（向量→文本）

把4中的结果转换成为文本

编辑文件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，上传到/home/hadoop/mahout/mahout_jar

如果需要清除eclipse中Launch Configuration中的信息，需要进入工程所在文件夹下的/.metadata/.plugins/org.eclipse.debug.core/.launches
然后删除里面的文件

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

这时候/output/canopy-output/part-m-00000里面放置的就是聚类的结果文件