Search in sources :

Example 1 with UpdatableSketch

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

the class UnionDoubleSummarySketchTest method exec.

@Test
public void exec() throws Exception {
    EvalFunc<Tuple> func = new UnionDoubleSummarySketch("4096");
    DataBag bag = BagFactory.getInstance().newDefaultBag();
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
        sketch.update(1, 1.0);
        sketch.update(2, 1.0);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
    }
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
        sketch.update(1, 1.0);
        sketch.update(2, 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);
    Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
    Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
    SketchIterator<DoubleSummary> it = sketch.iterator();
    while (it.next()) {
        Assert.assertEquals(it.getSummary().getValue(), 2.0, 0.0);
    }
}
Also used : UpdatableSketch(com.yahoo.sketches.tuple.UpdatableSketch) DataBag(org.apache.pig.data.DataBag) DoubleSummary(com.yahoo.sketches.tuple.DoubleSummary) UpdatableSketchBuilder(com.yahoo.sketches.tuple.UpdatableSketchBuilder) DoubleSummaryDeserializer(com.yahoo.sketches.tuple.DoubleSummaryDeserializer) DoubleSummaryFactory(com.yahoo.sketches.tuple.DoubleSummaryFactory) DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) Test(org.testng.annotations.Test)

Example 2 with UpdatableSketch

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

the class UnionDoubleSummarySketchTest method execMaxMode.

@Test
public void execMaxMode() throws Exception {
    EvalFunc<Tuple> func = new UnionDoubleSummarySketch("4096", "Max");
    DataBag bag = BagFactory.getInstance().newDefaultBag();
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
        sketch.update(1, 1.0);
        sketch.update(2, 1.0);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
    }
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
        sketch.update(1, 3.0);
        sketch.update(2, 3.0);
        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);
    Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
    Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
    SketchIterator<DoubleSummary> it = sketch.iterator();
    while (it.next()) {
        Assert.assertEquals(it.getSummary().getValue(), 3.0, 0.0);
    }
}
Also used : UpdatableSketch(com.yahoo.sketches.tuple.UpdatableSketch) DataBag(org.apache.pig.data.DataBag) DoubleSummary(com.yahoo.sketches.tuple.DoubleSummary) UpdatableSketchBuilder(com.yahoo.sketches.tuple.UpdatableSketchBuilder) DoubleSummaryDeserializer(com.yahoo.sketches.tuple.DoubleSummaryDeserializer) DoubleSummaryFactory(com.yahoo.sketches.tuple.DoubleSummaryFactory) DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) Test(org.testng.annotations.Test)

Example 3 with UpdatableSketch

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

the class UnionDoubleSummarySketchTest method algebraicIntemediateFinalExactMinMode.

@Test
public void algebraicIntemediateFinalExactMinMode() throws Exception {
    EvalFunc<Tuple> func = new UnionDoubleSummarySketch.IntermediateFinal("4096", "Min");
    DataBag bag = BagFactory.getInstance().newDefaultBag();
    // this is to simulate the output from Initial
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
        sketch.update(1, 1.0);
        sketch.update(2, 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
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
        sketch.update(1, 3.0);
        sketch.update(2, 3.0);
        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);
    Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
    Assert.assertEquals(sketch.getEstimate(), 2.0, 0.0);
    SketchIterator<DoubleSummary> it = sketch.iterator();
    while (it.next()) {
        Assert.assertEquals(it.getSummary().getValue(), 1.0, 0.0);
    }
}
Also used : UpdatableSketch(com.yahoo.sketches.tuple.UpdatableSketch) DataBag(org.apache.pig.data.DataBag) DoubleSummary(com.yahoo.sketches.tuple.DoubleSummary) UpdatableSketchBuilder(com.yahoo.sketches.tuple.UpdatableSketchBuilder) DoubleSummaryDeserializer(com.yahoo.sketches.tuple.DoubleSummaryDeserializer) DoubleSummaryFactory(com.yahoo.sketches.tuple.DoubleSummaryFactory) DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) Test(org.testng.annotations.Test)

Example 4 with UpdatableSketch

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

the class UnionDoubleSummarySketchTest method algebraicIntemediateFinalRandomized.

@Test
public void algebraicIntemediateFinalRandomized() throws Exception {
    EvalFunc<Tuple> func = new UnionDoubleSummarySketch.IntermediateFinal("16384");
    DataBag bag = BagFactory.getInstance().newDefaultBag();
    long key = 1;
    Random rnd = new Random();
    long uniques = 0;
    long updates = 0;
    // this is to simulate the output from a prior call of IntermediateFinal
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
        for (int i = 0; i < 40000; i++) sketch.update(key++, rnd.nextDouble() * 20);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
        uniques += 40000;
        updates += 40000;
    }
    // overlap
    key -= 20000;
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
        for (int i = 0; i < 60000; i++) sketch.update(key++, rnd.nextDouble() * 20);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
        uniques += 40000;
        updates += 60000;
    }
    // overlap
    key -= 20000;
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
        for (int i = 0; i < 60000; i++) sketch.update(key++, rnd.nextDouble() * 20);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
        uniques += 40000;
        updates += 60000;
    }
    // overlap
    key -= 20000;
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
        for (int i = 0; i < 60000; i++) sketch.update(key++, rnd.nextDouble() * 20);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
        uniques += 40000;
        updates += 60000;
    }
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
        for (int i = 0; i < 40000; i++) sketch.update(key++, rnd.nextDouble() * 20);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
        uniques += 40000;
        updates += 40000;
    }
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
        for (int i = 0; i < 40000; i++) sketch.update(key++, rnd.nextDouble() * 20);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
        uniques += 40000;
        updates += 40000;
    }
    // overlap
    key -= 20000;
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).setNominalEntries(16384).build();
        for (int i = 0; i < 60000; i++) sketch.update(key++, rnd.nextDouble() * 20);
        bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
        uniques += 40000;
        updates += 60000;
    }
    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);
    Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
    Assert.assertEquals(sketch.getEstimate(), uniques, uniques * 0.01);
    double sum = 0;
    SketchIterator<DoubleSummary> it = sketch.iterator();
    while (it.next()) {
        sum += it.getSummary().getValue();
    }
    // each update added 10 to the total on average
    // there is a slight chance of failing here
    Assert.assertEquals(sum / sketch.getTheta(), updates * 10.0, updates * 10.0 * 0.02);
}
Also used : DataBag(org.apache.pig.data.DataBag) DoubleSummaryFactory(com.yahoo.sketches.tuple.DoubleSummaryFactory) UpdatableSketch(com.yahoo.sketches.tuple.UpdatableSketch) Random(java.util.Random) DoubleSummary(com.yahoo.sketches.tuple.DoubleSummary) DoubleSummaryDeserializer(com.yahoo.sketches.tuple.DoubleSummaryDeserializer) DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) Test(org.testng.annotations.Test)

Example 5 with UpdatableSketch

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

the class UnionDoubleSummarySketchTest method accumulatorEmptySketch.

@Test
public void accumulatorEmptySketch() throws Exception {
    Accumulator<Tuple> func = new UnionDoubleSummarySketch("4096");
    DataBag bag = BagFactory.getInstance().newDefaultBag();
    {
        UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
        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);
    Sketch<DoubleSummary> sketch = Sketches.heapifySketch(Memory.wrap(bytes.get()), new DoubleSummaryDeserializer());
    Assert.assertEquals(sketch.getEstimate(), 0.0);
}
Also used : UpdatableSketch(com.yahoo.sketches.tuple.UpdatableSketch) DataBag(org.apache.pig.data.DataBag) DoubleSummary(com.yahoo.sketches.tuple.DoubleSummary) UpdatableSketchBuilder(com.yahoo.sketches.tuple.UpdatableSketchBuilder) DoubleSummaryDeserializer(com.yahoo.sketches.tuple.DoubleSummaryDeserializer) DoubleSummaryFactory(com.yahoo.sketches.tuple.DoubleSummaryFactory) DataByteArray(org.apache.pig.data.DataByteArray) Tuple(org.apache.pig.data.Tuple) Test(org.testng.annotations.Test)

Aggregations

UpdatableSketch (com.yahoo.sketches.tuple.UpdatableSketch)9 DataBag (org.apache.pig.data.DataBag)9 DataByteArray (org.apache.pig.data.DataByteArray)9 Tuple (org.apache.pig.data.Tuple)9 DoubleSummary (com.yahoo.sketches.tuple.DoubleSummary)8 DoubleSummaryDeserializer (com.yahoo.sketches.tuple.DoubleSummaryDeserializer)8 DoubleSummaryFactory (com.yahoo.sketches.tuple.DoubleSummaryFactory)8 Test (org.testng.annotations.Test)8 UpdatableSketchBuilder (com.yahoo.sketches.tuple.UpdatableSketchBuilder)5 DEFAULT_NOMINAL_ENTRIES (com.yahoo.sketches.Util.DEFAULT_NOMINAL_ENTRIES)1 Sketch (com.yahoo.sketches.tuple.Sketch)1 Union (com.yahoo.sketches.tuple.Union)1 Random (java.util.Random)1