1

我想找到我每个点的最近邻居，我使用 karlhigley ANN 模型进行了尝试。这是一段代码

List<Tuple2<Object, SparseVector>> svList = new ArrayList<>();
        svList.add(new Tuple2<Object, SparseVector>(3L,
                (Vectors.sparse(20, new int[] { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 },
                        new double[] { 5.0f, 3.0f, 4.0f, 5.0f, 5.0f, 1.0f, 5.0f, 3.0f, 4.0f, 5.0f, 5.0f, 3.0f, 4.0f,
                                5.0f, 5.0f, 1.0f, 5.0f, 3.0f, 4.0f, 5.0f })
                        .toSparse())));
        svList.add(new Tuple2<Object, SparseVector>(4L,
                (Vectors.sparse(20, new int[] { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 },
                        new double[] { 1.0f, 2.0f, 1.0f, 5.0f, 1.0f, 5.0f, 1.0f, 4.0f, 1.0f, 3.0, 1.0f, 2.0f, 1.0f,
                                5.0f, 1.0f, 5.0f, 1.0f, 4.0f, 1.0f, 3.0f })
                        .toSparse())));
        svList.add(new Tuple2<Object, SparseVector>(6L,
                (Vectors.sparse(20, new int[] { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 },
                        new double[] { 5.0f, 3.0f, 4.0f, 1.0f, 5.0f, 4.0f, 1.0f, 3.0f, 4.0f, 5.0f, 5.0f, 3.0f, 4.0f,
                                1.0f, 5.0f, 4.0f, 1.0f, 3.0f, 4.0f, 5.0f })
                        .toSparse())));

RDD<Tuple2<Object, SparseVector>> points = sc.parallelize(svList).rdd();

ANNModel annModel =
                new ANN(20, "cosine")
                .setTables(1)
                .setSignatureLength(20).setRandomSeed(3)
                .train(points,StorageLevel.MEMORY_AND_DISK());

JavaRDD<Tuple2<Object, Tuple2<Object, Object>[]>> neighbors2 = annModel.neighbors(3).toJavaRDD();

JavaRDD neighbors2 给了我所有的邻居和他们的分数为空。谁能帮助我了解我在哪里实施错误以及如何以正确的方式做到这一点？

这就是我打印输出的方式

neighbors2.foreach(f->{
            for(int i=0;i<f._2.length;i++){
                System.out.println(f._1+"====="+f._2[i]._1+"---"+f._2[i]._2);
            }
        });
4

1 回答 1

1

解决方案：

打印语句是错误的。它一定要是：

System.out.println(f._1()+"====="+f._2()[i]._1()+"---"+f._2()[i]._2());

原因：由于 f._2 和 f._2() 之间的类型转换问题。本帖将详细讲解： Java Tuple2使用访问器方法与直接调用变量的区别

于 2017-01-05T10:56:22.717 回答