Examples with GroupBy com.yahoo.bullet.query.aggregations.GroupBy used on opensource projects

Example 1 with GroupBy

use of com.yahoo.bullet.query.aggregations.GroupBy in project bullet-core by yahoo.

the class TupleSketchingStrategyTest method testCombining.

@Test
public void testCombining() {
    List<String> fields = asList("fieldA", "fieldB");
    TupleSketchingStrategy groupBy = makeGroupBy(fields, 5, new GroupOperation(COUNT, null, "count"), new GroupOperation(SUM, "price", "priceSum"));
    BulletRecord recordA = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("price", 3).getRecord();
    BulletRecord recordB = RecordBox.get().add("fieldA", "null").add("fieldB", "bar").add("price", 1).getRecord();
    IntStream.range(0, 10).mapToObj(i -> recordA).forEach(groupBy::consume);
    IntStream.range(0, 9).mapToObj(i -> recordB).forEach(groupBy::consume);
    IntStream.range(0, 20).mapToObj(i -> recordA).forEach(groupBy::consume);
    groupBy.consume(recordB);
    byte[] firstSerialized = groupBy.getData();
    // Remake it
    groupBy = makeGroupBy(fields, 5, new GroupOperation(COUNT, null, "count"), new GroupOperation(SUM, "price", "priceSum"));
    BulletRecord recordC = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("price", 3).getRecord();
    BulletRecord recordD = RecordBox.get().addNull("fieldA").addNull("fieldB").add("price", 10).getRecord();
    IntStream.range(0, 30).mapToObj(i -> recordC).forEach(groupBy::consume);
    IntStream.range(0, 10).mapToObj(i -> recordD).forEach(groupBy::consume);
    byte[] secondSerialized = groupBy.getData();
    groupBy = makeGroupBy(fields, 5, new GroupOperation(COUNT, null, "count"), new GroupOperation(SUM, "price", "priceSum"));
    groupBy.combine(firstSerialized);
    groupBy.combine(secondSerialized);
    Clip aggregate = groupBy.getResult();
    Assert.assertNotNull(aggregate);
    List<BulletRecord> records = aggregate.getRecords();
    Assert.assertEquals(records.size(), 3);
    // count = 10 + 20 + 30, price = 10*3 + 20*3 + 30*3
    BulletRecord expectedA = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("count", 60L).add("priceSum", 180.0).getRecord();
    // count = 9 + 1, price = 9*1 + 1*1
    BulletRecord expectedB = RecordBox.get().add("fieldA", "null").add("fieldB", "bar").add("count", 10L).add("priceSum", 10.0).getRecord();
    // count = 10, price = 10*10
    BulletRecord expectedC = RecordBox.get().add("fieldA", "null").add("fieldB", "null").add("count", 10L).add("priceSum", 100.0).getRecord();
    assertContains(records, expectedA);
    assertContains(records, expectedB);
    assertContains(records, expectedC);
    Assert.assertEquals(groupBy.getRecords(), aggregate.getRecords());
    Assert.assertEquals(groupBy.getMetadata().asMap(), aggregate.getMeta().asMap());
}

Example 2 with GroupBy

use of com.yahoo.bullet.query.aggregations.GroupBy in project bullet-core by yahoo.

the class TupleSketchingStrategyTest method testResetting.

@Test
public void testResetting() {
    List<String> fields = asList("fieldA", "fieldB");
    TupleSketchingStrategy groupBy = makeGroupBy(fields, 5, new GroupOperation(COUNT, null, "count"), new GroupOperation(SUM, "price", "priceSum"));
    BulletRecord recordA = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("price", 3).getRecord();
    BulletRecord recordB = RecordBox.get().add("fieldA", "null").add("fieldB", "bar").add("price", 1).getRecord();
    IntStream.range(0, 30).mapToObj(i -> recordA).forEach(groupBy::consume);
    IntStream.range(0, 10).mapToObj(i -> recordB).forEach(groupBy::consume);
    BulletRecord expectedA = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("count", 30L).add("priceSum", 90.0).getRecord();
    BulletRecord expectedB = RecordBox.get().add("fieldA", "null").add("fieldB", "bar").add("count", 10L).add("priceSum", 10.0).getRecord();
    Clip aggregate = groupBy.getResult();
    Assert.assertNotNull(aggregate);
    List<BulletRecord> records = aggregate.getRecords();
    Assert.assertEquals(records.size(), 2);
    assertContains(records, expectedA);
    assertContains(records, expectedB);
    Assert.assertEquals(groupBy.getRecords(), aggregate.getRecords());
    Assert.assertEquals(groupBy.getMetadata().asMap(), aggregate.getMeta().asMap());
    groupBy.reset();
    BulletRecord recordC = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("price", 3).getRecord();
    BulletRecord recordD = RecordBox.get().addNull("fieldA").addNull("fieldB").add("price", 10).getRecord();
    IntStream.range(0, 10).mapToObj(i -> recordB).forEach(groupBy::consume);
    IntStream.range(0, 30).mapToObj(i -> recordC).forEach(groupBy::consume);
    IntStream.range(0, 10).mapToObj(i -> recordD).forEach(groupBy::consume);
    aggregate = groupBy.getResult();
    Assert.assertNotNull(aggregate);
    records = aggregate.getRecords();
    Assert.assertEquals(records.size(), 3);
    expectedA = RecordBox.get().add("fieldA", "null").add("fieldB", "bar").add("count", 10L).add("priceSum", 10.0).getRecord();
    expectedB = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("count", 30L).add("priceSum", 90.0).getRecord();
    BulletRecord expectedC = RecordBox.get().add("fieldA", "null").add("fieldB", "null").add("count", 10L).add("priceSum", 100.0).getRecord();
    assertContains(records, expectedA);
    assertContains(records, expectedB);
    assertContains(records, expectedC);
    Assert.assertEquals(groupBy.getRecords(), aggregate.getRecords());
    Assert.assertEquals(groupBy.getMetadata().asMap(), aggregate.getMeta().asMap());
}

Example 3 with GroupBy

use of com.yahoo.bullet.query.aggregations.GroupBy in project bullet-core by yahoo.

the class TupleSketchingStrategyTest method testMoreGroupsThanNominalEntries.

@Test
public void testMoreGroupsThanNominalEntries() {
    Map<String, String> fields = singletonMap("fieldA", "A");
    // Nominal Entries is 32. Aggregation size is also 32
    TupleSketchingStrategy groupBy = makeGroupBy(makeConfiguration(32), fields, 32, singletonList(new GroupOperation(COUNT, null, "count")), ALL_METADATA);
    // Generate 4 batches of 64 records with 0 - 63 in fieldA.
    IntStream.range(0, 256).mapToObj(i -> RecordBox.get().add("fieldA", i % 64).getRecord()).forEach(groupBy::consume);
    Clip aggregate = groupBy.getResult();
    Assert.assertNotNull(aggregate);
    Map<String, Object> meta = aggregate.getMeta().asMap();
    Assert.assertEquals(meta.size(), 1);
    List<BulletRecord> records = aggregate.getRecords();
    Assert.assertEquals(records.size(), 32);
    Set<String> groups = new HashSet<>();
    for (BulletRecord record : records) {
        groups.add((String) record.typedGet("A").getValue());
        Assert.assertEquals(record.typedGet("count").getValue(), 4L);
    }
    Assert.assertEquals(groups.size(), 32);
    Assert.assertEquals(groupBy.getRecords(), aggregate.getRecords());
    Assert.assertEquals(groupBy.getMetadata().asMap(), aggregate.getMeta().asMap());
}

Example 4 with GroupBy

use of com.yahoo.bullet.query.aggregations.GroupBy in project bullet-core by yahoo.

the class TupleSketchingStrategyTest method testCombiningAndConsuming.

@Test
public void testCombiningAndConsuming() {
    List<String> fields = asList("fieldA", "fieldB");
    TupleSketchingStrategy groupBy = makeGroupBy(fields, 5, new GroupOperation(COUNT, null, "count"), new GroupOperation(SUM, "price", "priceSum"));
    BulletRecord recordA = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("price", 3).getRecord();
    BulletRecord recordB = RecordBox.get().add("fieldA", "null").add("fieldB", "bar").add("price", 1).getRecord();
    IntStream.range(0, 30).mapToObj(i -> recordA).forEach(groupBy::consume);
    IntStream.range(0, 10).mapToObj(i -> recordB).forEach(groupBy::consume);
    byte[] serialized = groupBy.getData();
    // Remake it
    groupBy = makeGroupBy(fields, 5, new GroupOperation(COUNT, null, "count"), new GroupOperation(SUM, "price", "priceSum"));
    BulletRecord recordC = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("price", 3).getRecord();
    BulletRecord recordD = RecordBox.get().addNull("fieldA").addNull("fieldB").add("price", 10).getRecord();
    IntStream.range(0, 30).mapToObj(i -> recordC).forEach(groupBy::consume);
    IntStream.range(0, 10).mapToObj(i -> recordD).forEach(groupBy::consume);
    groupBy.combine(serialized);
    Clip aggregate = groupBy.getResult();
    Assert.assertNotNull(aggregate);
    List<BulletRecord> records = aggregate.getRecords();
    Assert.assertEquals(records.size(), 3);
    BulletRecord expectedA = RecordBox.get().add("fieldA", "foo").add("fieldB", "bar").add("count", 60L).add("priceSum", 180.0).getRecord();
    BulletRecord expectedB = RecordBox.get().add("fieldA", "null").add("fieldB", "bar").add("count", 10L).add("priceSum", 10.0).getRecord();
    BulletRecord expectedC = RecordBox.get().add("fieldA", "null").add("fieldB", "null").add("count", 10L).add("priceSum", 100.0).getRecord();
    assertContains(records, expectedA);
    assertContains(records, expectedB);
    assertContains(records, expectedC);
    Assert.assertEquals(groupBy.getRecords(), aggregate.getRecords());
    Assert.assertEquals(groupBy.getMetadata().asMap(), aggregate.getMeta().asMap());
}

Example 5 with GroupBy

use of com.yahoo.bullet.query.aggregations.GroupBy in project bullet-core by yahoo.

the class TupleSketchingStrategyTest method testMetadata.

@Test
public void testMetadata() {
    Map<String, String> fields = singletonMap("fieldA", "fieldA");
    // Nominal Entries is 32. Aggregation size is also 32
    TupleSketchingStrategy groupBy = makeGroupBy(makeConfiguration(32), fields, 32, singletonList(new GroupOperation(COUNT, null, "count")), ALL_METADATA);
    // Generate 4 batches of 64 records with 0 - 63 in fieldA.
    IntStream.range(0, 256).mapToObj(i -> RecordBox.get().add("fieldA", i % 64).getRecord()).forEach(groupBy::consume);
    Clip aggregate = groupBy.getResult();
    Assert.assertNotNull(aggregate);
    List<BulletRecord> records = aggregate.getRecords();
    Assert.assertEquals(records.size(), 32);
    records.forEach(r -> Assert.assertTrue(Integer.valueOf((String) r.typedGet("fieldA").getValue()) < 64));
    Map<String, Object> meta = aggregate.getMeta().asMap();
    Assert.assertEquals(meta.size(), 1);
    Map<String, Object> stats = (Map<String, Object>) meta.get("aggregate_stats");
    Assert.assertEquals(stats.size(), 4);
    Assert.assertTrue((Boolean) stats.get("isEstimate"));
    double theta = (Double) stats.get("theta");
    Assert.assertTrue(theta <= 1.0);
    double groupEstimate = (Double) stats.get("uniquesApprox");
    Assert.assertTrue(stats.containsKey("stddev"));
    Map<String, Map<String, Double>> standardDeviations = (Map<String, Map<String, Double>>) stats.get("stddev");
    Assert.assertEquals(standardDeviations.size(), 3);
    double upperOneSigma = standardDeviations.get(KMVSketch.META_STD_DEV_1).get(KMVSketch.META_STD_DEV_UB);
    double lowerOneSigma = standardDeviations.get(KMVSketch.META_STD_DEV_1).get(KMVSketch.META_STD_DEV_LB);
    double upperTwoSigma = standardDeviations.get(KMVSketch.META_STD_DEV_2).get(KMVSketch.META_STD_DEV_UB);
    double lowerTwoSigma = standardDeviations.get(KMVSketch.META_STD_DEV_2).get(KMVSketch.META_STD_DEV_LB);
    double upperThreeSigma = standardDeviations.get(KMVSketch.META_STD_DEV_3).get(KMVSketch.META_STD_DEV_UB);
    double lowerThreeSigma = standardDeviations.get(KMVSketch.META_STD_DEV_3).get(KMVSketch.META_STD_DEV_LB);
    Assert.assertTrue(groupEstimate >= lowerOneSigma);
    Assert.assertTrue(groupEstimate <= upperOneSigma);
    Assert.assertTrue(groupEstimate >= lowerTwoSigma);
    Assert.assertTrue(groupEstimate <= upperTwoSigma);
    Assert.assertTrue(groupEstimate >= lowerThreeSigma);
    Assert.assertTrue(groupEstimate <= upperThreeSigma);
    Assert.assertTrue(groupEstimate <= upperThreeSigma);
    Assert.assertEquals(groupBy.getRecords(), aggregate.getRecords());
    Assert.assertEquals(groupBy.getMetadata().asMap(), aggregate.getMeta().asMap());
}