use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.
the class DataToArrayOfDoublesSketchTest method execEmptyBag.
@Test
public void execEmptyBag() throws Exception {
EvalFunc<Tuple> func = new DataToArrayOfDoublesSketch("1");
Tuple inputTuple = PigUtil.objectsToTuple(BagFactory.getInstance().newDefaultBag());
Tuple resultTuple = func.exec(inputTuple);
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);
}
use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.
the class DataToArrayOfDoublesSketchTest method execWithSampling.
@Test
public void execWithSampling() throws Exception {
EvalFunc<Tuple> func = new DataToArrayOfDoublesSketch("1024", "0.5", "1");
DataBag bag = BagFactory.getInstance().newDefaultBag();
int uniques = 10000;
for (int i = 0; i < uniques; i++) bag.add(PigUtil.objectsToTuple(i, 1.0));
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(), uniques, uniques * 0.01);
}
use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.
the class DataToArrayOfDoublesSketchTest method execAllInputTypes.
@Test
public void execAllInputTypes() throws Exception {
EvalFunc<Tuple> func = new DataToArrayOfDoublesSketch("32", "1");
DataBag bag = BagFactory.getInstance().newDefaultBag();
bag.add(PigUtil.objectsToTuple("a", 1.0));
bag.add(PigUtil.objectsToTuple("b", 1.0));
bag.add(PigUtil.objectsToTuple("a", 2.0));
bag.add(PigUtil.objectsToTuple("b", 2.0));
bag.add(PigUtil.objectsToTuple(1, 3.0));
bag.add(PigUtil.objectsToTuple(2L, 3.0));
bag.add(PigUtil.objectsToTuple(1f, 3.0));
bag.add(PigUtil.objectsToTuple(2.0, 3.0));
bag.add(PigUtil.objectsToTuple((byte) 3, 3.0));
bag.add(PigUtil.objectsToTuple(new DataByteArray("c".getBytes()), 3.0));
Tuple inputTuple = PigUtil.objectsToTuple(bag);
Tuple resultTuple = func.exec(inputTuple);
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(), 8.0, 0.0);
for (double[] values : sketch.getValues()) {
Assert.assertEquals(values[0], 3.0);
}
}
use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.
the class DataToArrayOfDoublesSketchTest method algebraicIntermediateFinal.
@Test
public void algebraicIntermediateFinal() throws Exception {
EvalFunc<Tuple> func = new DataToArrayOfDoublesSketch.IntermediateFinal("1");
Tuple inputTuple = TupleFactory.getInstance().newTuple(1);
DataBag bag = BagFactory.getInstance().newDefaultBag();
inputTuple.set(0, bag);
// this is to simulate the output from Initial
bag.add(PigUtil.objectsToTuple(PigUtil.tuplesToBag(PigUtil.objectsToTuple("a", 1.0))));
// this is to simulate the output from a prior call of IntermediateFinal
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().build();
sketch.update("b", new double[] { 1.0 });
sketch.update("a", new double[] { 2.0 });
sketch.update("b", new double[] { 2.0 });
bag.add(PigUtil.objectsToTuple(new DataByteArray(sketch.compact().toByteArray())));
}
Tuple resultTuple = func.exec(inputTuple);
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], 3.0);
}
}
use of com.yahoo.sketches.tuple.ArrayOfDoublesSketch in project sketches-pig by DataSketches.
the class UnionArrayOfDoublesSketchTest method accumulatorEmptySketch.
@Test
public void accumulatorEmptySketch() throws Exception {
Accumulator<Tuple> func = new UnionArrayOfDoublesSketch("1");
DataBag bag = BagFactory.getInstance().newDefaultBag();
{
ArrayOfDoublesUpdatableSketch sketch = new ArrayOfDoublesUpdatableSketchBuilder().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);
ArrayOfDoublesSketch sketch = ArrayOfDoublesSketches.heapifySketch(Memory.wrap(bytes.get()));
Assert.assertEquals(sketch.getEstimate(), 0.0);
}
Aggregations