我建议你用 Random 用于拆分数据集和 MultipleOutputs 将数据写入不同的路径。只需一个Map作业即可完成。下面是一个可以使用的Map器示例：

public class Splitter extends Mapper<LongWritable, Text, NullWritable, NullWritable> {
    MultipleOutputs mos;
    Random rnd = new Random();

    @Override
    protected void setup(Context context) throws IOException, InterruptedException {
        mos = new MultipleOutputs(context);
    }

    @Override
    protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
        if (rnd.nextDouble() < 0.8) {
            mos.write(key, value, "learning-set");
        } else {
            mos.write(key, value, "test-set");
        }
    }

    @Override
    protected void cleanup(Context context) throws IOException, InterruptedException {
        mos.close();
    }
}