@Test
public void exact() throws Exception {
EvalFunc<DataBag> func = new FrequentStringsSketchToEstimates();
ItemsSketch<String> sketch = new ItemsSketch<String>(8);
sketch.update("a");
sketch.update("a");
sketch.update("b");
Tuple inputTuple =
PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe())));
DataBag bag = func.exec(inputTuple);
Assert.assertNotNull(bag);
Assert.assertEquals(bag.size(), 2);
Iterator<Tuple> it = bag.iterator();
Tuple tuple1 = it.next();
Assert.assertEquals(tuple1.size(), 4);
Assert.assertEquals((String) tuple1.get(0), "a");
Assert.assertEquals((long) tuple1.get(1), 2L);
Assert.assertEquals((long) tuple1.get(2), 2L);
Assert.assertEquals((long) tuple1.get(3), 2L);
Tuple tuple2 = it.next();
Assert.assertEquals(tuple2.size(), 4);
Assert.assertEquals((String) tuple2.get(0), "b");
Assert.assertEquals((long) tuple2.get(1), 1L);
Assert.assertEquals((long) tuple2.get(2), 1L);
Assert.assertEquals((long) tuple2.get(3), 1L);
}
@Test
public void estimation() throws Exception {
ItemsSketch<String> sketch = new ItemsSketch<String>(8);
sketch.update("1", 1000);
sketch.update("2", 500);
sketch.update("3", 200);
sketch.update("4", 100);
sketch.update("5", 50);
sketch.update("6", 20);
sketch.update("7", 10);
sketch.update("8", 5);
sketch.update("9", 2);
sketch.update("10");
Tuple inputTuple =
PigUtil.objectsToTuple(new DataByteArray(sketch.toByteArray(new ArrayOfStringsSerDe())));
EvalFunc<DataBag> func1 = new FrequentStringsSketchToEstimates("NO_FALSE_POSITIVES");
DataBag bag1 = func1.exec(inputTuple);
Assert.assertNotNull(bag1);
Assert.assertTrue(bag1.size() < 10);
EvalFunc<DataBag> func2 = new FrequentStringsSketchToEstimates("NO_FALSE_NEGATIVES");
DataBag bag2 = func2.exec(inputTuple);
Assert.assertNotNull(bag2);
Assert.assertTrue(bag2.size() < 10);
Assert.assertTrue(bag1.size() < bag2.size());
}