MapReduce之計算平均數的Java與Python實現

Java代碼

• 重寫map：

public class WCMapper extends Mapper<LongWritable, Text, Text, Text> {

	@Override
	protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
		String line = value.toString();
		String[] words = line.split("\t");
		long sum = 0;
		long var = 0;
		long count = 0;
		
		// 計算讀進來讀每一行的數字個數與數字之和
		for (String word:words){

			int i = Integer.parseInt(word);
			sum += i;
			var += (i*i);
			count ++;

		}

		String sum0 = Long.toString(sum);
		String count0 = Long.toString(count);
		String var0 = Long.toString(var);
		String a = sum0+" "+count0+" "+var0;
		context.write(new Text("1"), new Text(a));

	}


}

• 重寫reduc：

public class WCReducer extends Reducer<Text, Text, Text, Text> {
	@Override
	protected void reduce(Text word, Iterable<Text> value, Context context) throws IOException, InterruptedException {

		Double sum = 0.0;
		long count = 0;
		BigInteger var = new BigInteger("0");
		// 由於關鍵字都是"1"，所以一次reduce能讀入所有的數
		// 計算所有數字之和與數字的個數之和
		for (Text one:value){
			String line = one.toString();
			String[] words = line.split(" ");
			Double i = Double.valueOf(words[0]);
			sum += i;
			long j = Long.parseLong(words[1]);
			count += j;
			long m = Long.parseLong(words[2]);
			BigInteger k = BigInteger.valueOf(m);
			var = var.add(k);
		}
		// 計算平均數
		String avg = String.valueOf (sum/count);
		//double var0 = (double) (var/count) - (sum*sum);
		context.write(new Text("平均數爲："), new Text(avg));
		//context.write(new Text("方差爲："), new DoubleWritable(var0));
	}
}

• 構造主類

public class WCDriver {

	public static void main(String[] args) throws IOException, ClassNotFoundException, InterruptedException {

		Configuration conf = new Configuration();
//		conf.set("mapreduce.task.io.sort.factor", "100");
		Job job = Job.getInstance(conf);
		
		job.setJarByClass(WCDriver.class);

		job.setMapOutputKeyClass(Text.class);
		job.setMapOutputValueClass(Text.class);
		job.setOutputKeyClass(Text.class);
		job.setOutputValueClass(DoubleWritable.class);

		job.setMapperClass(WCMapper.class);
		job.setReducerClass(WCReducer.class);
		
		//job.setCombinerClass(myReducer.class);
		
		job.setNumReduceTasks(1);

		FileInputFormat.setInputPaths(job, new Path(args[0]));
		FileOutputFormat.setOutputPath(job, new Path(args[1]));
		

		boolean bool = job.waitForCompletion(true);
		System.exit(bool?0:1);

	}

}

Python代碼：

• mapper：

#!/usr/bin/python
# -*- coding: utf-8 -*-

import sys


# 在map端拿到當前map所有讀入數據的數字之和與數字個數
def map():
    count = 0
    sum = 0
    for line in sys.stdin:
        line = line.strip()
        words = line.split()
        for word in words:
            sum += int(word)
            count += 1
    print('%d\t%d' % (sum, count))


if __name__ == '__main__':
    map()

• reducer：

#!/usr/bin/python
# -*- coding: utf-8 -*-

import sys


def reduce():
    sum = 0
    count = 0
    for line in sys.stdin:
        sum_count = line.strip().split('\t')
        sum += int(sum_count[0])
        count += int(sum_count[1])
    a = sum/count
    print('%s\t%s' % ("平均數：", str(a)))


if __name__ == '__main__':
    reduce()