Search in sources :

Example 21 with ArrayOfDoublesSketch

use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.

the class UnionArrayOfDoublesSketchTest method accumulator.

@Test
public void accumulator() throws Exception {
    Accumulator<Tuple> func = new UnionArrayOfDoublesSketch("4096", "1");
    DataBag bag = BagFactory.getInstance().newDefaultBag();
    {
        ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().build();
        sketch.update(1, new double[] { 1.0 });
        sketch.update(2, new double[] { 1.0 });
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
    }
    func.accumulate(PigUtil.objectsToTuple(bag));
    bag = BagFactory.getInstance().newDefaultBag();
    {
        ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().build();
        sketch.update(1, new double[] { 1.0 });
        sketch.update(2, new double[] { 1.0 });
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
    }
    func.accumulate(PigUtil.objectsToTuple(bag));
    Tuple resultTuple = func.getValue();
    Assert.assertNotNull(resultTuple);
    Assert.assertEquals(resultTuple.size(), 1);
    DataByteArray bytes = (DataByteArray) resultTuple.get(0);
    Assert.assertTrue(bytes.size() > 0);
    ArrayOfDoublesSketch sketch = ArrayOfDoublesSketches.heapifySketch(Memory.wrap(bytes.get()));
    Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
    for (double[] values : sketch.getValues()) {
        Assert.assertEquals(values[0], 2.0, 0.0);
    }
}
Also used : ArrayOfDoublesUpdatableSketchBuilder(com.yahoo.sketches.tuple.ArrayOfDoublesUpdatableSketchBuilder) DataBag(org.apache.pig.data.DataBag) ArrayOfDoublesUpdatableSketch(com.yahoo.sketches.tuple.ArrayOfDoublesUpdatableSketch) DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) ArrayOfDoublesSketch(com.yahoo.sketches.tuple.ArrayOfDoublesSketch) Test(org.testng.annotations.Test)

Example 22 with ArrayOfDoublesSketch

use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.

the class UnionArrayOfDoublesSketchTest method accumulatorEmptyInputTuple.

@Test
public void accumulatorEmptyInputTuple() throws Exception {
    Accumulator<Tuple> func = new UnionArrayOfDoublesSketch();
    func.accumulate(TupleFactory.getInstance().newTuple());
    Tuple resultTuple = func.getValue();
    Assert.assertNotNull(resultTuple);
    Assert.assertEquals(resultTuple.size(), 1);
    DataByteArray bytes = (DataByteArray) resultTuple.get(0);
    Assert.assertTrue(bytes.size() > 0);
    ArrayOfDoublesSketch sketch = ArrayOfDoublesSketches.heapifySketch(Memory.wrap(bytes.get()));
    Assert.assertEquals(sketch.getEstimate(), 0.0);
}
Also used : DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) ArrayOfDoublesSketch(com.yahoo.sketches.tuple.ArrayOfDoublesSketch) Test(org.testng.annotations.Test)

Example 23 with ArrayOfDoublesSketch

use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.

the class UnionArrayOfDoublesSketchTest method algebraicIntemediateFinalEstimation.

@Test
public void algebraicIntemediateFinalEstimation() throws Exception {
    EvalFunc<Tuple> func = new UnionArrayOfDoublesSketch.IntermediateFinal("16384", "1");
    DataBag bag = BagFactory.getInstance().newDefaultBag();
    long value = 1;
    // this is to simulate the output from Initial
    {
        ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
        for (int i = 0; i < 20000; i++) sketch.update(value++, new double[] { 1.0 });
        DataBag innerBag = PigUtil.tuplesToBag(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
        bag.add(PigUtil.objectsToTuple(innerBag));
    }
    // this is to simulate the output from a prior call of IntermediateFinal
    {
        ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().setNominalEntries(16384).build();
        for (int i = 0; i < 20000; i++) sketch.update(value++, new double[] { 1.0 });
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
    }
    Tuple resultTuple = func.exec(PigUtil.objectsToTuple(bag));
    Assert.assertNotNull(resultTuple);
    Assert.assertEquals(resultTuple.size(), 1);
    DataByteArray bytes = (DataByteArray) resultTuple.get(0);
    Assert.assertTrue(bytes.size() > 0);
    ArrayOfDoublesSketch sketch = ArrayOfDoublesSketches.heapifySketch(Memory.wrap(bytes.get()));
    Assert.assertEquals(sketch.getEstimate(), 40000.0, 40000.0 * 0.01);
    for (double[] values : sketch.getValues()) {
        Assert.assertEquals(values[0], 1.0, 0.0);
    }
}
Also used : ArrayOfDoublesUpdatableSketchBuilder(com.yahoo.sketches.tuple.ArrayOfDoublesUpdatableSketchBuilder) DataBag(org.apache.pig.data.DataBag) ArrayOfDoublesUpdatableSketch(com.yahoo.sketches.tuple.ArrayOfDoublesUpdatableSketch) DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) ArrayOfDoublesSketch(com.yahoo.sketches.tuple.ArrayOfDoublesSketch) Test(org.testng.annotations.Test)

Example 24 with ArrayOfDoublesSketch

use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.

the class UnionArrayOfDoublesSketchTest method accumulatorNullInput.

@Test
public void accumulatorNullInput() throws Exception {
    Accumulator<Tuple> func = new UnionArrayOfDoublesSketch();
    func.accumulate(null);
    Tuple resultTuple = func.getValue();
    Assert.assertNotNull(resultTuple);
    Assert.assertEquals(resultTuple.size(), 1);
    DataByteArray bytes = (DataByteArray) resultTuple.get(0);
    Assert.assertTrue(bytes.size() > 0);
    ArrayOfDoublesSketch sketch = ArrayOfDoublesSketches.heapifySketch(Memory.wrap(bytes.get()));
    Assert.assertEquals(sketch.getEstimate(), 0.0);
}
Also used : DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) ArrayOfDoublesSketch(com.yahoo.sketches.tuple.ArrayOfDoublesSketch) Test(org.testng.annotations.Test)

Example 25 with ArrayOfDoublesSketch

use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.

the class UnionArrayOfDoublesSketchTest method accumulatorEmptyInnerTuple.

@Test
public void accumulatorEmptyInnerTuple() throws Exception {
    Accumulator<Tuple> func = new UnionArrayOfDoublesSketch();
    func.accumulate(PigUtil.objectsToTuple(PigUtil.tuplesToBag(TupleFactory.getInstance().newTuple())));
    Tuple resultTuple = func.getValue();
    Assert.assertNotNull(resultTuple);
    Assert.assertEquals(resultTuple.size(), 1);
    DataByteArray bytes = (DataByteArray) resultTuple.get(0);
    Assert.assertTrue(bytes.size() > 0);
    ArrayOfDoublesSketch sketch = ArrayOfDoublesSketches.heapifySketch(Memory.wrap(bytes.get()));
    Assert.assertEquals(sketch.getEstimate(), 0.0);
}
Also used : DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) ArrayOfDoublesSketch(com.yahoo.sketches.tuple.ArrayOfDoublesSketch) Test(org.testng.annotations.Test)

Aggregations

ArrayOfDoublesSketch (com.yahoo.sketches.tuple.ArrayOfDoublesSketch)26 DataByteArray (org.apache.pig.data.DataByteArray)26 Tuple (org.apache.pig.data.Tuple)22 Test (org.testng.annotations.Test)19 DataBag (org.apache.pig.data.DataBag)12 ArrayOfDoublesUpdatableSketch (com.yahoo.sketches.tuple.ArrayOfDoublesUpdatableSketch)8 ArrayOfDoublesUpdatableSketchBuilder (com.yahoo.sketches.tuple.ArrayOfDoublesUpdatableSketchBuilder)8 SummaryStatistics (org.apache.commons.math3.stat.descriptive.SummaryStatistics)3 ArrayOfDoublesSketchIterator (com.yahoo.sketches.tuple.ArrayOfDoublesSketchIterator)2 DoublesSketchBuilder (com.yahoo.sketches.quantiles.DoublesSketchBuilder)1 UpdateDoublesSketch (com.yahoo.sketches.quantiles.UpdateDoublesSketch)1 Random (java.util.Random)1 TTest (org.apache.commons.math3.stat.inference.TTest)1