今天就跟大家聊聊有关如何理解TopK算法及其实现，可能很多人都不太了解，为了让大家更加了解，小编给大家总结了以下内容，希望大家根据这篇文章可以有所收获。
1、问题描述
在大数据规模中，经常遇到一类需要求出现频率最高的K个数，这类问题称为“TOPK”问题！例如：统计歌曲中最热门的前10首歌曲，统计访问流量最高的前5个网站等。
2、例如统计访问流量最高的前5个网站：
数据test.data文件：
数据格式解释：域名 上行流量 下行流量
思路：
1、Mapper每解析一行内容,按照"\t"获取各个字段
2、因为URL有很多重复记录,所以将URL放到key(通过分析MapReduce原理),流量放在value
3、在reduce统计总流量,通过TreeMap进行对数据进行缓存,最后一并输出(值得注意的是要一次性输出必须要用到Reduce类的cleanup方法)
程序如下：
Mapper类：
package com.itheima.hadoop.mapreduce.mapper;
import java.io.IOException;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Counter;
import com.itheima.hadoop.mapreduce.bean.FlowBean;
public class TopKURLMapper extends Mapper<LongWritable, Text, Text, FlowBean> {
/**
* @param key
* : 每一行偏移量
* @param value
* : 每一行的内容
* @param context
* : 环境上下文
*/
@Override
public void map(LongWritable key, Text value, Context context)
throws IOException, InterruptedException {
/**
* 该计数器是org.apache.hadoop.mapreduce.Counter
*/
Counter counter = context
.getCounter("ExistProblem", "ExistProblemLine"); // 自定义存在问题的行错误计数器
String line = value.toString(); // 读取一行数据
String[] fields = line.split("\t"); // 获取各个字段,按照\t划分
try {
String url = fields[0]; // 获取URL字段
long upFlow = Long.parseLong(fields[1]); // 获取上行流量(upFlow)字段
long downFlow = Long.parseLong(fields[2]); // 获取下行流量(downFlow)字段
FlowBean bean = new FlowBean(upFlow, downFlow); // 将上行流量和下行流量封装到bean中
Text tUrl = new Text(url); // 将java数据类型转换hadoop数据类型
context.write(tUrl, bean); // 传递的数据较多，封装到bean进行传输(tips:bean传输时需要注意序列化问题)
} catch (Exception e) {
e.printStackTrace();
counter.increment(1); // 记录错误行数
}
}
}
Reduce类：
package com.itheima.hadoop.mapreduce.reducer;
import java.io.IOException;
import java.util.Map.Entry;
import java.util.TreeMap;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Reducer;
import com.itheima.hadoop.mapreduce.bean.FlowBean;
public class TopKURLReducer extends Reducer<Text, FlowBean, FlowBean, Text> {
private TreeMap<FlowBean, Text> treeMap = new TreeMap<FlowBean, Text>();
/**
* @param key
* : 每一行相同URL
* @param values
* : 总流量bean
*/
@Override
public void reduce(Text key, Iterable<FlowBean> values, Context context)
throws IOException, InterruptedException {
long countUpFlow = 0;
long countDownFlow = 0;
/*
* 1、取出每个bean的总流量 2、统计多个bean的总流量 3、缓存到treeMap中
*/
for (FlowBean bean : values) {
countUpFlow += bean.getUpFlow(); // 统计上行流量
countDownFlow += bean.getDownFlow(); // 统计下行总流量
}
// 封装统计的流量
FlowBean bean = new FlowBean(countUpFlow, countDownFlow);
treeMap.put(bean, new Text(key)); // 缓存到treeMap中
}
@Override
public void cleanup(Context context) throws IOException,
InterruptedException {
//遍历缓存
for (Entry<FlowBean,Text> entry : treeMap.entrySet()) {
context.write(entry.getKey(), entry.getValue());
}
super.cleanup(context); // 不能动原本的销毁操作
}
}
FlowBean类：
package com.itheima.hadoop.mapreduce.bean;
import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
import org.apache.hadoop.io.Writable;
public class FlowBean implements Writable, Comparable<FlowBean> {
private long upFlow;
private long downFlow;
private long maxFlow;
@Override
public String toString() {
return upFlow + "\t" + downFlow + "\t" + maxFlow;
}
/**
* 1、序列化注意的问题,序列化需要默认的构造方法(反射) 2、在readFields()和write()方法中,应该遵循按照顺序写出和读入
*/
public FlowBean() {
}
public FlowBean(long upFlow, long downFlow) {
this.upFlow = upFlow;
this.downFlow = downFlow;
this.maxFlow = upFlow + downFlow;
}
public long getUpFlow() {
return upFlow;
}
public void setUpFlow(long upFlow) {
this.upFlow = upFlow;
}
public long getDownFlow() {
return downFlow;
}
public void setDownFlow(long downFlow) {
this.downFlow = downFlow;
}
public long getMaxFlow() {
return maxFlow;
}
public void setMaxFlow(long maxFlow) {
this.maxFlow = maxFlow;
}
@Override
public void readFields(DataInput dataIn) throws IOException {
upFlow = dataIn.readLong();
downFlow = dataIn.readLong();
maxFlow = dataIn.readLong();
}
@Override
public void write(DataOutput dataOut) throws IOException {
dataOut.writeLong(upFlow);
dataOut.writeLong(downFlow);
dataOut.writeLong(maxFlow);
}
@Override
public int compareTo(FlowBean o) {
return this.maxFlow > o.maxFlow ? -1
: this.maxFlow < o.maxFlow ? 1 : 0;
}
}
驱动类：
package com.itheima.hadoop.drivers;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.Tool;
import com.itheima.hadoop.mapreduce.bean.FlowBean;
import com.itheima.hadoop.mapreduce.mapper.TopKURLMapper;
import com.itheima.hadoop.mapreduce.reducer.TopKURLReducer;
public class TopKURLDriver extends Configured implements Tool{
@Override
public int run(String[] args) throws Exception {
/**
* 1、创建job作业
* 2、设置job提交的Class
* 3、设置MapperClass,设置ReduceClass
* 4、设置Mapper和Reduce各自的OutputKey和OutputValue类型
* 5、设置处理文件的路径,输出结果的路径
* 6、提交job
*/
Configuration conf = new Configuration();
Job job = Job.getInstance(conf);
job.setJarByClass(TopKURLRunner.class);
job.setMapperClass(TopKURLMapper.class);
job.setReducerClass(TopKURLReducer.class);
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(FlowBean.class);
job.setOutputKeyClass(FlowBean.class);
job.setOutputValueClass(Text.class);
FileInputFormat.setInputPaths(job, new Path(args[0]));
FileOutputFormat.setOutputPath(job,new Path(args[1]));
//参数true为打印进度
return job.waitForCompletion(true)?0:1;
}
}
package com.itheima.hadoop.runner;
import org.apache.hadoop.util.ToolRunner;
import com.itheima.hadoop.runner.TopKURLRunner;
public class TopKURLRunner {
public static void main(String[] args) throws Exception {
int res = ToolRunner.run(new TopKURLRunner(), args);
System.exit(res);
}
}
运行命令：hadoop jar topkurl.jar com.itheima.hadoop.drives.TopKURLDriver /test/inputData /test/outputData
运行结果：
支付宝扫一扫
微信扫一扫
.png)
.png)
.png)
.png)
