org.apache.spark.ml.linalg.VectorUDT Java Examples

public static Dataset<Row> binaryBlockToDataFrame(SparkSession sparkSession,
                                                  JavaPairRDD<MatrixIndexes, MatrixBlock> in, DataCharacteristics mc, boolean toVector)
{
	if( !mc.colsKnown() )
		throw new RuntimeException("Number of columns needed to convert binary block to data frame.");
	
	//slice blocks into rows, align and convert into data frame rows
	JavaRDD<Row> rowsRDD = in
		.flatMapToPair(new SliceBinaryBlockToRowsFunction(mc.getBlocksize()))
		.groupByKey().map(new ConvertRowBlocksToRows((int)mc.getCols(), mc.getBlocksize(), toVector));
	
	//create data frame schema
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(DF_ID_COLUMN, DataTypes.DoubleType, false));
	if( toVector )
		fields.add(DataTypes.createStructField("C1", new VectorUDT(), false));
	else { // row
		for(int i = 1; i <= mc.getCols(); i++)
			fields.add(DataTypes.createStructField("C"+i, DataTypes.DoubleType, false));
	}
	
	//rdd to data frame conversion
	return sparkSession.createDataFrame(rowsRDD.rdd(), DataTypes.createStructType(fields));
}
 

public static Dataset<Row> binaryBlockToDataFrame(SparkSession sparkSession,
                                                  JavaPairRDD<MatrixIndexes, MatrixBlock> in, DataCharacteristics mc, boolean toVector)
{
	if( !mc.colsKnown() )
		throw new RuntimeException("Number of columns needed to convert binary block to data frame.");
	
	//slice blocks into rows, align and convert into data frame rows
	JavaRDD<Row> rowsRDD = in
		.flatMapToPair(new SliceBinaryBlockToRowsFunction(mc.getBlocksize()))
		.groupByKey().map(new ConvertRowBlocksToRows((int)mc.getCols(), mc.getBlocksize(), toVector));
	
	//create data frame schema
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(DF_ID_COLUMN, DataTypes.DoubleType, false));
	if( toVector )
		fields.add(DataTypes.createStructField("C1", new VectorUDT(), false));
	else { // row
		for(int i = 1; i <= mc.getCols(); i++)
			fields.add(DataTypes.createStructField("C"+i, DataTypes.DoubleType, false));
	}
	
	//rdd to data frame conversion
	return sparkSession.createDataFrame(rowsRDD.rdd(), DataTypes.createStructType(fields));
}
 

@Test
public void testDataFrameSumDMLVectorWithIDColumn() {
	System.out.println("MLContextTest - DataFrame sum DML, vector with ID column");

	List<Tuple2<Double, Vector>> list = new ArrayList<>();
	list.add(new Tuple2<>(1.0, Vectors.dense(1.0, 2.0, 3.0)));
	list.add(new Tuple2<>(2.0, Vectors.dense(4.0, 5.0, 6.0)));
	list.add(new Tuple2<>(3.0, Vectors.dense(7.0, 8.0, 9.0)));
	JavaRDD<Tuple2<Double, Vector>> javaRddTuple = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddTuple.map(new DoubleVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(RDDConverterUtils.DF_ID_COLUMN, DataTypes.DoubleType, true));
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	MatrixMetadata mm = new MatrixMetadata(MatrixFormat.DF_VECTOR_WITH_INDEX);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame, mm);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLMllibVectorWithIDColumn() {
	System.out.println("MLContextTest - DataFrame sum DML, mllib vector with ID column");

	List<Tuple2<Double, org.apache.spark.mllib.linalg.Vector>> list = new ArrayList<>();
	list.add(new Tuple2<>(1.0, org.apache.spark.mllib.linalg.Vectors.dense(1.0, 2.0, 3.0)));
	list.add(new Tuple2<>(2.0, org.apache.spark.mllib.linalg.Vectors.dense(4.0, 5.0, 6.0)));
	list.add(new Tuple2<>(3.0, org.apache.spark.mllib.linalg.Vectors.dense(7.0, 8.0, 9.0)));
	JavaRDD<Tuple2<Double, org.apache.spark.mllib.linalg.Vector>> javaRddTuple = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddTuple.map(new DoubleMllibVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(RDDConverterUtils.DF_ID_COLUMN, DataTypes.DoubleType, true));
	fields.add(DataTypes.createStructField("C1", new org.apache.spark.mllib.linalg.VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	MatrixMetadata mm = new MatrixMetadata(MatrixFormat.DF_VECTOR_WITH_INDEX);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame, mm);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLVectorWithNoIDColumn() {
	System.out.println("MLContextTest - DataFrame sum DML, vector with no ID column");

	List<Vector> list = new ArrayList<>();
	list.add(Vectors.dense(1.0, 2.0, 3.0));
	list.add(Vectors.dense(4.0, 5.0, 6.0));
	list.add(Vectors.dense(7.0, 8.0, 9.0));
	JavaRDD<Vector> javaRddVector = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddVector.map(new VectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	MatrixMetadata mm = new MatrixMetadata(MatrixFormat.DF_VECTOR);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame, mm);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLMllibVectorWithNoIDColumn() {
	System.out.println("MLContextTest - DataFrame sum DML, mllib vector with no ID column");

	List<org.apache.spark.mllib.linalg.Vector> list = new ArrayList<>();
	list.add(org.apache.spark.mllib.linalg.Vectors.dense(1.0, 2.0, 3.0));
	list.add(org.apache.spark.mllib.linalg.Vectors.dense(4.0, 5.0, 6.0));
	list.add(org.apache.spark.mllib.linalg.Vectors.dense(7.0, 8.0, 9.0));
	JavaRDD<org.apache.spark.mllib.linalg.Vector> javaRddVector = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddVector.map(new MllibVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField("C1", new org.apache.spark.mllib.linalg.VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	MatrixMetadata mm = new MatrixMetadata(MatrixFormat.DF_VECTOR);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame, mm);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLVectorWithIDColumnNoFormatSpecified() {
	System.out.println("MLContextTest - DataFrame sum DML, vector with ID column, no format specified");

	List<Tuple2<Double, Vector>> list = new ArrayList<>();
	list.add(new Tuple2<>(1.0, Vectors.dense(1.0, 2.0, 3.0)));
	list.add(new Tuple2<>(2.0, Vectors.dense(4.0, 5.0, 6.0)));
	list.add(new Tuple2<>(3.0, Vectors.dense(7.0, 8.0, 9.0)));
	JavaRDD<Tuple2<Double, Vector>> javaRddTuple = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddTuple.map(new DoubleVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(RDDConverterUtils.DF_ID_COLUMN, DataTypes.DoubleType, true));
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLVectorWithNoIDColumnNoFormatSpecified() {
	System.out.println("MLContextTest - DataFrame sum DML, vector with no ID column, no format specified");

	List<Vector> list = new ArrayList<>();
	list.add(Vectors.dense(1.0, 2.0, 3.0));
	list.add(Vectors.dense(4.0, 5.0, 6.0));
	list.add(Vectors.dense(7.0, 8.0, 9.0));
	JavaRDD<Vector> javaRddVector = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddVector.map(new VectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumPYDMLVectorWithNoIDColumnNoFormatSpecified() {
	System.out.println("MLContextTest - DataFrame sum PYDML, vector with no ID column, no format specified");

	List<Vector> list = new ArrayList<>();
	list.add(Vectors.dense(1.0, 2.0, 3.0));
	list.add(Vectors.dense(4.0, 5.0, 6.0));
	list.add(Vectors.dense(7.0, 8.0, 9.0));
	JavaRDD<Vector> javaRddVector = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddVector.map(new VectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	Script script = dml("print('sum: ' + sum(M))").in("M", dataFrame);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLVectorWithNoIDColumnNoFormatSpecified() {
	System.out.println("MLContextTest - DataFrame sum DML, vector with no ID column, no format specified");

	List<Vector> list = new ArrayList<>();
	list.add(Vectors.dense(1.0, 2.0, 3.0));
	list.add(Vectors.dense(4.0, 5.0, 6.0));
	list.add(Vectors.dense(7.0, 8.0, 9.0));
	JavaRDD<Vector> javaRddVector = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddVector.map(new VectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumPYDMLVectorWithIDColumnNoFormatSpecified() {
	System.out.println("MLContextTest - DataFrame sum PYDML, vector with ID column, no format specified");

	List<Tuple2<Double, Vector>> list = new ArrayList<>();
	list.add(new Tuple2<>(1.0, Vectors.dense(1.0, 2.0, 3.0)));
	list.add(new Tuple2<>(2.0, Vectors.dense(4.0, 5.0, 6.0)));
	list.add(new Tuple2<>(3.0, Vectors.dense(7.0, 8.0, 9.0)));
	JavaRDD<Tuple2<Double, Vector>> javaRddTuple = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddTuple.map(new DoubleVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(RDDConverterUtils.DF_ID_COLUMN, DataTypes.DoubleType, true));
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	Script script = dml("print('sum: ' + sum(M))").in("M", dataFrame);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLVectorWithIDColumnNoFormatSpecified() {
	System.out.println("MLContextTest - DataFrame sum DML, vector with ID column, no format specified");

	List<Tuple2<Double, Vector>> list = new ArrayList<>();
	list.add(new Tuple2<>(1.0, Vectors.dense(1.0, 2.0, 3.0)));
	list.add(new Tuple2<>(2.0, Vectors.dense(4.0, 5.0, 6.0)));
	list.add(new Tuple2<>(3.0, Vectors.dense(7.0, 8.0, 9.0)));
	JavaRDD<Tuple2<Double, Vector>> javaRddTuple = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddTuple.map(new DoubleVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(RDDConverterUtils.DF_ID_COLUMN, DataTypes.DoubleType, true));
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLMllibVectorWithNoIDColumn() {
	System.out.println("MLContextTest - DataFrame sum DML, mllib vector with no ID column");

	List<org.apache.spark.mllib.linalg.Vector> list = new ArrayList<>();
	list.add(org.apache.spark.mllib.linalg.Vectors.dense(1.0, 2.0, 3.0));
	list.add(org.apache.spark.mllib.linalg.Vectors.dense(4.0, 5.0, 6.0));
	list.add(org.apache.spark.mllib.linalg.Vectors.dense(7.0, 8.0, 9.0));
	JavaRDD<org.apache.spark.mllib.linalg.Vector> javaRddVector = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddVector.map(new MllibVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField("C1", new org.apache.spark.mllib.linalg.VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	MatrixMetadata mm = new MatrixMetadata(MatrixFormat.DF_VECTOR);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame, mm);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLVectorWithNoIDColumn() {
	System.out.println("MLContextTest - DataFrame sum DML, vector with no ID column");

	List<Vector> list = new ArrayList<>();
	list.add(Vectors.dense(1.0, 2.0, 3.0));
	list.add(Vectors.dense(4.0, 5.0, 6.0));
	list.add(Vectors.dense(7.0, 8.0, 9.0));
	JavaRDD<Vector> javaRddVector = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddVector.map(new VectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	MatrixMetadata mm = new MatrixMetadata(MatrixFormat.DF_VECTOR);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame, mm);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLMllibVectorWithIDColumn() {
	System.out.println("MLContextTest - DataFrame sum DML, mllib vector with ID column");

	List<Tuple2<Double, org.apache.spark.mllib.linalg.Vector>> list = new ArrayList<>();
	list.add(new Tuple2<>(1.0, org.apache.spark.mllib.linalg.Vectors.dense(1.0, 2.0, 3.0)));
	list.add(new Tuple2<>(2.0, org.apache.spark.mllib.linalg.Vectors.dense(4.0, 5.0, 6.0)));
	list.add(new Tuple2<>(3.0, org.apache.spark.mllib.linalg.Vectors.dense(7.0, 8.0, 9.0)));
	JavaRDD<Tuple2<Double, org.apache.spark.mllib.linalg.Vector>> javaRddTuple = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddTuple.map(new DoubleMllibVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(RDDConverterUtils.DF_ID_COLUMN, DataTypes.DoubleType, true));
	fields.add(DataTypes.createStructField("C1", new org.apache.spark.mllib.linalg.VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	MatrixMetadata mm = new MatrixMetadata(MatrixFormat.DF_VECTOR_WITH_INDEX);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame, mm);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

@Test
public void testDataFrameSumDMLVectorWithIDColumn() {
	System.out.println("MLContextTest - DataFrame sum DML, vector with ID column");

	List<Tuple2<Double, Vector>> list = new ArrayList<>();
	list.add(new Tuple2<>(1.0, Vectors.dense(1.0, 2.0, 3.0)));
	list.add(new Tuple2<>(2.0, Vectors.dense(4.0, 5.0, 6.0)));
	list.add(new Tuple2<>(3.0, Vectors.dense(7.0, 8.0, 9.0)));
	JavaRDD<Tuple2<Double, Vector>> javaRddTuple = sc.parallelize(list);

	JavaRDD<Row> javaRddRow = javaRddTuple.map(new DoubleVectorRow());
	List<StructField> fields = new ArrayList<>();
	fields.add(DataTypes.createStructField(RDDConverterUtils.DF_ID_COLUMN, DataTypes.DoubleType, true));
	fields.add(DataTypes.createStructField("C1", new VectorUDT(), true));
	StructType schema = DataTypes.createStructType(fields);
	Dataset<Row> dataFrame = spark.createDataFrame(javaRddRow, schema);

	MatrixMetadata mm = new MatrixMetadata(MatrixFormat.DF_VECTOR_WITH_INDEX);

	Script script = dml("print('sum: ' + sum(M));").in("M", dataFrame, mm);
	setExpectedStdOut("sum: 45.0");
	ml.execute(script);
}
 

/**
 * If the MatrixFormat of the DataFrame has not been explicitly specified,
 * attempt to determine the proper MatrixFormat.
 *
 * @param dataFrame
 *            the Spark {@code DataFrame}
 * @param matrixMetadata
 *            the matrix metadata, if available
 */
public static void determineMatrixFormatIfNeeded(Dataset<Row> dataFrame, MatrixMetadata matrixMetadata) {
	if (matrixMetadata == null) {
		return;
	}
	MatrixFormat matrixFormat = matrixMetadata.getMatrixFormat();
	if (matrixFormat != null) {
		return;
	}
	StructType schema = dataFrame.schema();
	boolean hasID = false;
	try {
		schema.fieldIndex(RDDConverterUtils.DF_ID_COLUMN);
		hasID = true;
	} catch (IllegalArgumentException iae) {
	}

	StructField[] fields = schema.fields();
	MatrixFormat mf = null;
	if (hasID) {
		if (fields[1].dataType() instanceof VectorUDT) {
			mf = MatrixFormat.DF_VECTOR_WITH_INDEX;
		} else {
			mf = MatrixFormat.DF_DOUBLES_WITH_INDEX;
		}
	} else {
		if (fields[0].dataType() instanceof VectorUDT) {
			mf = MatrixFormat.DF_VECTOR;
		} else {
			mf = MatrixFormat.DF_DOUBLES;
		}
	}

	if (mf == null) {
		throw new MLContextException("DataFrame format not recognized as an accepted SystemDS MatrixFormat");
	}
	matrixMetadata.setMatrixFormat(mf);
}
 

/**
 * Obtain column vector from DataFrame schema
 * 
 * @param dfschema schema as StructType
 * @param containsID if true, contains ID column
 * @return 0-based column index of vector column, -1 if no vector.
 */
private static int getColVectFromDFSchema(StructType dfschema, boolean containsID) {
	int off = containsID ? 1 : 0;
	for( int i=off; i<dfschema.fields().length; i++ ) {
		StructField structType = dfschema.apply(i);
		if(structType.dataType() instanceof VectorUDT)
			return i-off;
	}
	
	return -1;
}
 

@Test
public void testOutputDataFrameOfVectorsDML() {
	System.out.println("MLContextTest - output DataFrame of vectors DML");

	String s = "m=matrix('1 2 3 4',rows=2,cols=2);";
	Script script = dml(s).out("m");
	MLResults results = ml.execute(script);
	Dataset<Row> df = results.getDataFrame("m", true);
	Dataset<Row> sortedDF = df.sort(RDDConverterUtils.DF_ID_COLUMN);

	// verify column types
	StructType schema = sortedDF.schema();
	StructField[] fields = schema.fields();
	StructField idColumn = fields[0];
	StructField vectorColumn = fields[1];
	Assert.assertTrue(idColumn.dataType() instanceof DoubleType);
	Assert.assertTrue(vectorColumn.dataType() instanceof VectorUDT);

	List<Row> list = sortedDF.collectAsList();

	Row row1 = list.get(0);
	Assert.assertEquals(1.0, row1.getDouble(0), 0.0);
	Vector v1 = (DenseVector) row1.get(1);
	double[] arr1 = v1.toArray();
	Assert.assertArrayEquals(new double[] { 1.0, 2.0 }, arr1, 0.0);

	Row row2 = list.get(1);
	Assert.assertEquals(2.0, row2.getDouble(0), 0.0);
	Vector v2 = (DenseVector) row2.get(1);
	double[] arr2 = v2.toArray();
	Assert.assertArrayEquals(new double[] { 3.0, 4.0 }, arr2, 0.0);
}
 

/**
 * If the MatrixFormat of the DataFrame has not been explicitly specified,
 * attempt to determine the proper MatrixFormat.
 *
 * @param dataFrame
 *            the Spark {@code DataFrame}
 * @param matrixMetadata
 *            the matrix metadata, if available
 */
public static void determineMatrixFormatIfNeeded(Dataset<Row> dataFrame, MatrixMetadata matrixMetadata) {
	if (matrixMetadata == null) {
		return;
	}
	MatrixFormat matrixFormat = matrixMetadata.getMatrixFormat();
	if (matrixFormat != null) {
		return;
	}
	StructType schema = dataFrame.schema();
	boolean hasID = false;
	try {
		schema.fieldIndex(RDDConverterUtils.DF_ID_COLUMN);
		hasID = true;
	} catch (IllegalArgumentException iae) {
	}

	StructField[] fields = schema.fields();
	MatrixFormat mf = null;
	if (hasID) {
		if (fields[1].dataType() instanceof VectorUDT) {
			mf = MatrixFormat.DF_VECTOR_WITH_INDEX;
		} else {
			mf = MatrixFormat.DF_DOUBLES_WITH_INDEX;
		}
	} else {
		if (fields[0].dataType() instanceof VectorUDT) {
			mf = MatrixFormat.DF_VECTOR;
		} else {
			mf = MatrixFormat.DF_DOUBLES;
		}
	}

	if (mf == null) {
		throw new MLContextException("DataFrame format not recognized as an accepted SystemDS MatrixFormat");
	}
	matrixMetadata.setMatrixFormat(mf);
}
 

@Test
public void testMinMaxScaler() {
    //prepare data
    JavaRDD<Row> jrdd = jsc.parallelize(Arrays.asList(
            RowFactory.create(1.0, Vectors.dense(data[0])),
            RowFactory.create(2.0, Vectors.dense(data[1])),
            RowFactory.create(3.0, Vectors.dense(data[2])),
            RowFactory.create(4.0, Vectors.dense(data[3]))
    ));

    StructType schema = new StructType(new StructField[]{
            new StructField("label", DataTypes.DoubleType, false, Metadata.empty()),
            new StructField("features", new VectorUDT(), false, Metadata.empty())
    });

    Dataset<Row> df = spark.createDataFrame(jrdd, schema);

    //train model in spark
    MinMaxScalerModel sparkModel = new MinMaxScaler()
            .setInputCol("features")
            .setOutputCol("scaled")
            .setMin(-5)
            .setMax(5)
            .fit(df);


    //Export model, import it back and get transformer
    byte[] exportedModel = ModelExporter.export(sparkModel);
    final Transformer transformer = ModelImporter.importAndGetTransformer(exportedModel);

    //compare predictions
    List<Row> sparkOutput = sparkModel.transform(df).orderBy("label").select("features", "scaled").collectAsList();
    assertCorrectness(sparkOutput, expected, transformer);
}
 

private void start() {
  SparkSession spark = SparkSession.builder().appName(
      "Simple prediction from Text File").master("local").getOrCreate();

  spark.udf().register("vectorBuilder", new VectorBuilder(), new VectorUDT());

  String filename = "data/tuple-data-file.csv";
  StructType schema = new StructType(
      new StructField[] { new StructField("_c0", DataTypes.DoubleType, false,
          Metadata.empty()),
          new StructField("_c1", DataTypes.DoubleType, false, Metadata
              .empty()),
          new StructField("features", new VectorUDT(), true, Metadata
              .empty()), });

  Dataset<Row> df = spark.read().format("csv").schema(schema).option("header",
      "false")
      .load(filename);
  df = df.withColumn("valuefeatures", df.col("_c0")).drop("_c0");
  df = df.withColumn("label", df.col("_c1")).drop("_c1");
  df.printSchema();

  df = df.withColumn("features", callUDF("vectorBuilder", df.col(
      "valuefeatures")));
  df.printSchema();
  df.show();

  LinearRegression lr = new LinearRegression().setMaxIter(20);// .setRegParam(1).setElasticNetParam(1);

  // Fit the model to the data.
  LinearRegressionModel model = lr.fit(df);

  // Given a dataset, predict each point's label, and show the results.
  model.transform(df).show();

  LinearRegressionTrainingSummary trainingSummary = model.summary();
  System.out.println("numIterations: " + trainingSummary.totalIterations());
  System.out.println("objectiveHistory: " + Vectors.dense(trainingSummary
      .objectiveHistory()));
  trainingSummary.residuals().show();
  System.out.println("RMSE: " + trainingSummary.rootMeanSquaredError());
  System.out.println("r2: " + trainingSummary.r2());

  double intercept = model.intercept();
  System.out.println("Interesection: " + intercept);
  double regParam = model.getRegParam();
  System.out.println("Regression parameter: " + regParam);
  double tol = model.getTol();
  System.out.println("Tol: " + tol);
  Double feature = 7.0;
  Vector features = Vectors.dense(feature);
  double p = model.predict(features);

  System.out.println("Prediction for feature " + feature + " is " + p);
  System.out.println(8 * regParam + intercept);
}
 

public static void main(String[] args) {
  SparkSession spark = SparkSession
    .builder()
    .appName("JavaElementwiseProductExample")
    .getOrCreate();

  // $example on$
  // Create some vector data; also works for sparse vectors
  List<Row> data = Arrays.asList(
    RowFactory.create("a", Vectors.dense(1.0, 2.0, 3.0)),
    RowFactory.create("b", Vectors.dense(4.0, 5.0, 6.0))
  );

  List<StructField> fields = new ArrayList<StructField>(2);
  fields.add(DataTypes.createStructField("id", DataTypes.StringType, false));
  fields.add(DataTypes.createStructField("vector", new VectorUDT(), false));

  StructType schema = DataTypes.createStructType(fields);

  Dataset<Row> dataFrame = spark.createDataFrame(data, schema);

  Vector transformingVector = Vectors.dense(0.0, 1.0, 2.0);

  ElementwiseProduct transformer = new ElementwiseProduct()
    .setScalingVec(transformingVector)
    .setInputCol("vector")
    .setOutputCol("transformedVector");

  // Batch transform the vectors to create new column:
  transformer.transform(dataFrame).show();
  // $example off$
  spark.stop();
}
 

public static void main(String[] args) {
  SparkSession spark = SparkSession
    .builder()
    .appName("JavaPolynomialExpansionExample")
    .getOrCreate();

  // $example on$
  PolynomialExpansion polyExpansion = new PolynomialExpansion()
    .setInputCol("features")
    .setOutputCol("polyFeatures")
    .setDegree(3);

  List<Row> data = Arrays.asList(
    RowFactory.create(Vectors.dense(2.0, 1.0)),
    RowFactory.create(Vectors.dense(0.0, 0.0)),
    RowFactory.create(Vectors.dense(3.0, -1.0))
  );
  StructType schema = new StructType(new StructField[]{
    new StructField("features", new VectorUDT(), false, Metadata.empty()),
  });
  Dataset<Row> df = spark.createDataFrame(data, schema);

  Dataset<Row> polyDF = polyExpansion.transform(df);
  polyDF.show(false);
  // $example off$

  spark.stop();
}
 

public static void main(String[] args) {
  SparkSession spark = SparkSession
    .builder()
    .appName("JavaAFTSurvivalRegressionExample")
    .getOrCreate();

  // $example on$
  List<Row> data = Arrays.asList(
    RowFactory.create(1.218, 1.0, Vectors.dense(1.560, -0.605)),
    RowFactory.create(2.949, 0.0, Vectors.dense(0.346, 2.158)),
    RowFactory.create(3.627, 0.0, Vectors.dense(1.380, 0.231)),
    RowFactory.create(0.273, 1.0, Vectors.dense(0.520, 1.151)),
    RowFactory.create(4.199, 0.0, Vectors.dense(0.795, -0.226))
  );
  StructType schema = new StructType(new StructField[]{
    new StructField("label", DataTypes.DoubleType, false, Metadata.empty()),
    new StructField("censor", DataTypes.DoubleType, false, Metadata.empty()),
    new StructField("features", new VectorUDT(), false, Metadata.empty())
  });
  Dataset<Row> training = spark.createDataFrame(data, schema);
  double[] quantileProbabilities = new double[]{0.3, 0.6};
  AFTSurvivalRegression aft = new AFTSurvivalRegression()
    .setQuantileProbabilities(quantileProbabilities)
    .setQuantilesCol("quantiles");

  AFTSurvivalRegressionModel model = aft.fit(training);

  // Print the coefficients, intercept and scale parameter for AFT survival regression
  System.out.println("Coefficients: " + model.coefficients());
  System.out.println("Intercept: " + model.intercept());
  System.out.println("Scale: " + model.scale());
  model.transform(training).show(false);
  // $example off$

  spark.stop();
}
 

public static void main(String[] args) {
  SparkSession spark = SparkSession
    .builder()
    .appName("JavaNormalizerExample")
    .getOrCreate();

  // $example on$
  List<Row> data = Arrays.asList(
      RowFactory.create(0, Vectors.dense(1.0, 0.1, -8.0)),
      RowFactory.create(1, Vectors.dense(2.0, 1.0, -4.0)),
      RowFactory.create(2, Vectors.dense(4.0, 10.0, 8.0))
  );
  StructType schema = new StructType(new StructField[]{
      new StructField("id", DataTypes.IntegerType, false, Metadata.empty()),
      new StructField("features", new VectorUDT(), false, Metadata.empty())
  });
  Dataset<Row> dataFrame = spark.createDataFrame(data, schema);

  // Normalize each Vector using $L^1$ norm.
  Normalizer normalizer = new Normalizer()
    .setInputCol("features")
    .setOutputCol("normFeatures")
    .setP(1.0);

  Dataset<Row> l1NormData = normalizer.transform(dataFrame);
  l1NormData.show();

  // Normalize each Vector using $L^\infty$ norm.
  Dataset<Row> lInfNormData =
    normalizer.transform(dataFrame, normalizer.p().w(Double.POSITIVE_INFINITY));
  lInfNormData.show();
  // $example off$

  spark.stop();
}
 

/**
 * Obtain column vector from DataFrame schema
 * 
 * @param dfschema schema as StructType
 * @param containsID if true, contains ID column
 * @return 0-based column index of vector column, -1 if no vector.
 */
private static int getColVectFromDFSchema(StructType dfschema, boolean containsID) {
	int off = containsID ? 1 : 0;
	for( int i=off; i<dfschema.fields().length; i++ ) {
		StructField structType = dfschema.apply(i);
		if(structType.dataType() instanceof VectorUDT)
			return i-off;
	}
	
	return -1;
}
 

@Test
public void testOutputDataFrameOfVectorsDML() {
	System.out.println("MLContextTest - output DataFrame of vectors DML");

	String s = "m=matrix('1 2 3 4',rows=2,cols=2);";
	Script script = dml(s).out("m");
	MLResults results = ml.execute(script);
	Dataset<Row> df = results.getDataFrame("m", true);
	Dataset<Row> sortedDF = df.sort(RDDConverterUtils.DF_ID_COLUMN);

	// verify column types
	StructType schema = sortedDF.schema();
	StructField[] fields = schema.fields();
	StructField idColumn = fields[0];
	StructField vectorColumn = fields[1];
	Assert.assertTrue(idColumn.dataType() instanceof DoubleType);
	Assert.assertTrue(vectorColumn.dataType() instanceof VectorUDT);

	List<Row> list = sortedDF.collectAsList();

	Row row1 = list.get(0);
	Assert.assertEquals(1.0, row1.getDouble(0), 0.0);
	Vector v1 = (DenseVector) row1.get(1);
	double[] arr1 = v1.toArray();
	Assert.assertArrayEquals(new double[] { 1.0, 2.0 }, arr1, 0.0);

	Row row2 = list.get(1);
	Assert.assertEquals(2.0, row2.getDouble(0), 0.0);
	Vector v2 = (DenseVector) row2.get(1);
	double[] arr2 = v2.toArray();
	Assert.assertArrayEquals(new double[] { 3.0, 4.0 }, arr2, 0.0);
}
 

/**
 * One-hot encodes a protein sequence. The one-hot encoding
    * encodes the 20 natural amino acids, plus X for any other 
    * residue for a total of 21 elements per residue.
 * 
 * @return dataset with feature vector appended
 */
public Dataset<Row> oneHotEncode() {
	SparkSession session = data.sparkSession();
	int maxLength = getMaxSequenceLength(data);

	session.udf().register("encoder", new UDF1<String, Vector>() {
		private static final long serialVersionUID = -6095318836772114908L;

		@Override
		public Vector call(String s) throws Exception {
			int len = AMINO_ACIDS21.size();
               double[] values = new double[len * maxLength];
			char[] seq = s.toCharArray();
			for (int i = 0; i < seq.length; i++) {
				int index = AMINO_ACIDS21.indexOf(seq[i]);
				// replace any non-matching code, e.g., U, with X
				if (index == -1) {
					index = AMINO_ACIDS21.indexOf('X');
				}
				values[i * len + index] = 1;
			}

			return Vectors.dense(values);
		}
	}, new VectorUDT());

	// append feature column
	data.createOrReplaceTempView("table");
	data = session.sql("SELECT *, encoder(" 
	+ inputCol + ") AS " 
			+ outputCol + " from table");
	
	return data;
}
 

/**
 * Encodes a protein sequence by 7 physicochemical
 * properties. 
 * 
 * <p> See:  Meiler, J., Müller, M., Zeidler, A. et al. J Mol Model (2001) 7: 360. doi:
 * <a href="https://link.springer.com/article/10.1007/s008940100038">10.1007/s008940100038</a>
    *
 * @return dataset with feature vector appended
 */
public Dataset<Row> propertyEncode() {
	SparkSession session = data.sparkSession();
    int maxLength = getMaxSequenceLength(data);

	session.udf().register("encoder", new UDF1<String, Vector>(){
		private static final long serialVersionUID = 1L;

		@Override
		public Vector call(String s) throws Exception {
               double[] values = new double[7*maxLength];
			for (int i = 0, k = 0; i < s.length(); i++) {
				double[] property = properties.get(s.charAt(i));
				if (property != null) {
					for (double p: property) {
						values[k++] = p;
					}
				}	
			}
			return Vectors.dense(values);
		}
	}, new VectorUDT());

	// append feature column
			data.createOrReplaceTempView("table");
			data = session.sql("SELECT *, encoder(" 
			+ inputCol + ") AS " 
					+ outputCol + " from table");
			
			return data;
}