use of com.yahoo.sketches.tuple.DoubleSummaryFactory in project sketches-pig by DataSketches.
the class DoubleSummarySketchToEstimatesTest method normalCase.
@Test
public void normalCase() throws Exception {
EvalFunc<Tuple> func = new DoubleSummarySketchToEstimates();
UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
int iterations = 100000;
for (int i = 0; i < iterations; i++) sketch.update(i, 1.0);
for (int i = 0; i < iterations; i++) sketch.update(i, 1.0);
Tuple inputTuple = PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray()));
Tuple resultTuple = func.exec(inputTuple);
Assert.assertNotNull(resultTuple);
Assert.assertEquals(resultTuple.size(), 2);
Assert.assertEquals((double) resultTuple.get(0), iterations, iterations * 0.03);
Assert.assertEquals((double) resultTuple.get(1), 2 * iterations, 2 * iterations * 0.03);
}
use of com.yahoo.sketches.tuple.DoubleSummaryFactory in project sketches-pig by DataSketches.
the class DoubleSummarySketchToPercentileTest method emptySketch.
@Test
public void emptySketch() throws Exception {
EvalFunc<Double> func = new DoubleSummarySketchToPercentile();
UpdatableSketch<Double, DoubleSummary> sketch = new UpdatableSketchBuilder<Double, DoubleSummary>(new DoubleSummaryFactory()).build();
Tuple inputTuple = TupleFactory.getInstance().newTuple(Arrays.asList(new DataByteArray(sketch.compact().toByteArray()), 0.0));
double result = func.exec(inputTuple);
Assert.assertEquals(result, Double.NaN);
}
use of com.yahoo.sketches.tuple.DoubleSummaryFactory 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);
}
}
use of com.yahoo.sketches.tuple.DoubleSummaryFactory 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);
}
}
use of com.yahoo.sketches.tuple.DoubleSummaryFactory 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);
}
}
Aggregations