org.apache.flink.api.java.summarize.BooleanColumnSummary Java Examples

The following examples show how to use org.apache.flink.api.java.summarize.BooleanColumnSummary. You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example #1
Source File: BooleanValueSummaryAggregatorTest.java    From flink with Apache License 2.0 6 votes vote down vote up
/**
 * Helper method for summarizing a list of values.
 *
 * <p>This method breaks the rule of "testing only one thing" by aggregating and combining
 * a bunch of different ways.
 */
protected BooleanColumnSummary summarize(Boolean... values) {

	BooleanValue[] booleanValues = new BooleanValue[values.length];
	for (int i = 0; i < values.length; i++) {
		if (values[i] != null) {
			booleanValues[i] = new BooleanValue(values[i]);
		}
	}

	return new AggregateCombineHarness<BooleanValue, BooleanColumnSummary, ValueSummaryAggregator.BooleanValueSummaryAggregator>() {
		@Override
		protected void compareResults(BooleanColumnSummary result1, BooleanColumnSummary result2) {
			Assert.assertEquals(result1.getNullCount(), result2.getNullCount());
			Assert.assertEquals(result1.getNonNullCount(), result2.getNonNullCount());
			Assert.assertEquals(result1.getTrueCount(), result2.getTrueCount());
			Assert.assertEquals(result1.getFalseCount(), result2.getFalseCount());
		}
	}.summarize(booleanValues);
}
 
Example #2
Source File: BooleanValueSummaryAggregatorTest.java    From Flink-CEPplus with Apache License 2.0 6 votes vote down vote up
/**
 * Helper method for summarizing a list of values.
 *
 * <p>This method breaks the rule of "testing only one thing" by aggregating and combining
 * a bunch of different ways.
 */
protected BooleanColumnSummary summarize(Boolean... values) {

	BooleanValue[] booleanValues = new BooleanValue[values.length];
	for (int i = 0; i < values.length; i++) {
		if (values[i] != null) {
			booleanValues[i] = new BooleanValue(values[i]);
		}
	}

	return new AggregateCombineHarness<BooleanValue, BooleanColumnSummary, ValueSummaryAggregator.BooleanValueSummaryAggregator>() {
		@Override
		protected void compareResults(BooleanColumnSummary result1, BooleanColumnSummary result2) {
			Assert.assertEquals(result1.getNullCount(), result2.getNullCount());
			Assert.assertEquals(result1.getNonNullCount(), result2.getNonNullCount());
			Assert.assertEquals(result1.getTrueCount(), result2.getTrueCount());
			Assert.assertEquals(result1.getFalseCount(), result2.getFalseCount());
		}
	}.summarize(booleanValues);
}
 
Example #3
Source File: BooleanValueSummaryAggregatorTest.java    From flink with Apache License 2.0 6 votes vote down vote up
/**
 * Helper method for summarizing a list of values.
 *
 * <p>This method breaks the rule of "testing only one thing" by aggregating and combining
 * a bunch of different ways.
 */
protected BooleanColumnSummary summarize(Boolean... values) {

	BooleanValue[] booleanValues = new BooleanValue[values.length];
	for (int i = 0; i < values.length; i++) {
		if (values[i] != null) {
			booleanValues[i] = new BooleanValue(values[i]);
		}
	}

	return new AggregateCombineHarness<BooleanValue, BooleanColumnSummary, ValueSummaryAggregator.BooleanValueSummaryAggregator>() {
		@Override
		protected void compareResults(BooleanColumnSummary result1, BooleanColumnSummary result2) {
			Assert.assertEquals(result1.getNullCount(), result2.getNullCount());
			Assert.assertEquals(result1.getNonNullCount(), result2.getNonNullCount());
			Assert.assertEquals(result1.getTrueCount(), result2.getTrueCount());
			Assert.assertEquals(result1.getFalseCount(), result2.getFalseCount());
		}
	}.summarize(booleanValues);
}
 
Example #4
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllNullBooleans() {
	BooleanColumnSummary summary = summarize(null, null, null, null);
	Assert.assertEquals(4, summary.getTotalCount());
	Assert.assertEquals(4, summary.getNullCount());
	Assert.assertEquals(0, summary.getNonNullCount());
	Assert.assertEquals(0, summary.getTrueCount());
	Assert.assertEquals(0, summary.getFalseCount());
}
 
Example #5
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
/**
 * Helper method for summarizing a list of values.
 *
 * <p>This method breaks the rule of "testing only one thing" by aggregating and combining
 * a bunch of different ways.
 */
protected BooleanColumnSummary summarize(Boolean... values) {
	return new AggregateCombineHarness<Boolean, BooleanColumnSummary, BooleanSummaryAggregator>() {
		@Override
		protected void compareResults(BooleanColumnSummary result1, BooleanColumnSummary result2) {
			Assert.assertEquals(result1.getNullCount(), result2.getNullCount());
			Assert.assertEquals(result1.getNonNullCount(), result2.getNonNullCount());
			Assert.assertEquals(result1.getTrueCount(), result2.getTrueCount());
			Assert.assertEquals(result1.getFalseCount(), result2.getFalseCount());
		}
	}.summarize(values);
}
 
Example #6
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllFalse() {
	BooleanColumnSummary summary = summarize(false, false, false);
	Assert.assertEquals(3, summary.getTotalCount());
	Assert.assertEquals(0, summary.getNullCount());
	Assert.assertEquals(3, summary.getNonNullCount());
	Assert.assertEquals(0, summary.getTrueCount());
	Assert.assertEquals(3, summary.getFalseCount());
}
 
Example #7
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllTrue() {
	BooleanColumnSummary summary = summarize(true, true, true, true, true, true);
	Assert.assertEquals(6, summary.getTotalCount());
	Assert.assertEquals(0, summary.getNullCount());
	Assert.assertEquals(6, summary.getNonNullCount());
	Assert.assertEquals(6, summary.getTrueCount());
	Assert.assertEquals(0, summary.getFalseCount());
}
 
Example #8
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllNullBooleans() {
	BooleanColumnSummary summary = summarize(null, null, null, null);
	Assert.assertEquals(4, summary.getTotalCount());
	Assert.assertEquals(4, summary.getNullCount());
	Assert.assertEquals(0, summary.getNonNullCount());
	Assert.assertEquals(0, summary.getTrueCount());
	Assert.assertEquals(0, summary.getFalseCount());
}
 
Example #9
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
@Test
public void testMixedGroup() {
	BooleanColumnSummary summary = summarize(true, false, null, true, true, true, false, null, true, false, true);
	Assert.assertEquals(11, summary.getTotalCount());
	Assert.assertEquals(2, summary.getNullCount());
	Assert.assertEquals(9, summary.getNonNullCount());
	Assert.assertEquals(6, summary.getTrueCount());
	Assert.assertEquals(3, summary.getFalseCount());
}
 
Example #10
Source File: BooleanSummaryAggregator.java    From flink with Apache License 2.0 5 votes vote down vote up
@Override
public void combine(Aggregator<Boolean, BooleanColumnSummary> otherSameType) {
	BooleanSummaryAggregator other = (BooleanSummaryAggregator) otherSameType;
	trueCount += other.trueCount;
	falseCount += other.falseCount;
	nullCount += other.nullCount;
}
 
Example #11
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
/**
 * Helper method for summarizing a list of values.
 *
 * <p>This method breaks the rule of "testing only one thing" by aggregating and combining
 * a bunch of different ways.
 */
protected BooleanColumnSummary summarize(Boolean... values) {
	return new AggregateCombineHarness<Boolean, BooleanColumnSummary, BooleanSummaryAggregator>() {
		@Override
		protected void compareResults(BooleanColumnSummary result1, BooleanColumnSummary result2) {
			Assert.assertEquals(result1.getNullCount(), result2.getNullCount());
			Assert.assertEquals(result1.getNonNullCount(), result2.getNonNullCount());
			Assert.assertEquals(result1.getTrueCount(), result2.getTrueCount());
			Assert.assertEquals(result1.getFalseCount(), result2.getFalseCount());
		}
	}.summarize(values);
}
 
Example #12
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllFalse() {
	BooleanColumnSummary summary = summarize(false, false, false);
	Assert.assertEquals(3, summary.getTotalCount());
	Assert.assertEquals(0, summary.getNullCount());
	Assert.assertEquals(3, summary.getNonNullCount());
	Assert.assertEquals(0, summary.getTrueCount());
	Assert.assertEquals(3, summary.getFalseCount());
}
 
Example #13
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllTrue() {
	BooleanColumnSummary summary = summarize(true, true, true, true, true, true);
	Assert.assertEquals(6, summary.getTotalCount());
	Assert.assertEquals(0, summary.getNullCount());
	Assert.assertEquals(6, summary.getNonNullCount());
	Assert.assertEquals(6, summary.getTrueCount());
	Assert.assertEquals(0, summary.getFalseCount());
}
 
Example #14
Source File: BooleanSummaryAggregatorTest.java    From flink with Apache License 2.0 5 votes vote down vote up
@Test
public void testMixedGroup() {
	BooleanColumnSummary summary = summarize(true, false, null, true, true, true, false, null, true, false, true);
	Assert.assertEquals(11, summary.getTotalCount());
	Assert.assertEquals(2, summary.getNullCount());
	Assert.assertEquals(9, summary.getNonNullCount());
	Assert.assertEquals(6, summary.getTrueCount());
	Assert.assertEquals(3, summary.getFalseCount());
}
 
Example #15
Source File: BooleanSummaryAggregator.java    From flink with Apache License 2.0 5 votes vote down vote up
@Override
public void combine(Aggregator<Boolean, BooleanColumnSummary> otherSameType) {
	BooleanSummaryAggregator other = (BooleanSummaryAggregator) otherSameType;
	trueCount += other.trueCount;
	falseCount += other.falseCount;
	nullCount += other.nullCount;
}
 
Example #16
Source File: BooleanSummaryAggregatorTest.java    From Flink-CEPplus with Apache License 2.0 5 votes vote down vote up
/**
 * Helper method for summarizing a list of values.
 *
 * <p>This method breaks the rule of "testing only one thing" by aggregating and combining
 * a bunch of different ways.
 */
protected BooleanColumnSummary summarize(Boolean... values) {
	return new AggregateCombineHarness<Boolean, BooleanColumnSummary, BooleanSummaryAggregator>() {
		@Override
		protected void compareResults(BooleanColumnSummary result1, BooleanColumnSummary result2) {
			Assert.assertEquals(result1.getNullCount(), result2.getNullCount());
			Assert.assertEquals(result1.getNonNullCount(), result2.getNonNullCount());
			Assert.assertEquals(result1.getTrueCount(), result2.getTrueCount());
			Assert.assertEquals(result1.getFalseCount(), result2.getFalseCount());
		}
	}.summarize(values);
}
 
Example #17
Source File: BooleanSummaryAggregatorTest.java    From Flink-CEPplus with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllFalse() {
	BooleanColumnSummary summary = summarize(false, false, false);
	Assert.assertEquals(3, summary.getTotalCount());
	Assert.assertEquals(0, summary.getNullCount());
	Assert.assertEquals(3, summary.getNonNullCount());
	Assert.assertEquals(0, summary.getTrueCount());
	Assert.assertEquals(3, summary.getFalseCount());
}
 
Example #18
Source File: BooleanSummaryAggregatorTest.java    From Flink-CEPplus with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllTrue() {
	BooleanColumnSummary summary = summarize(true, true, true, true, true, true);
	Assert.assertEquals(6, summary.getTotalCount());
	Assert.assertEquals(0, summary.getNullCount());
	Assert.assertEquals(6, summary.getNonNullCount());
	Assert.assertEquals(6, summary.getTrueCount());
	Assert.assertEquals(0, summary.getFalseCount());
}
 
Example #19
Source File: BooleanSummaryAggregatorTest.java    From Flink-CEPplus with Apache License 2.0 5 votes vote down vote up
@Test
public void testAllNullBooleans() {
	BooleanColumnSummary summary = summarize(null, null, null, null);
	Assert.assertEquals(4, summary.getTotalCount());
	Assert.assertEquals(4, summary.getNullCount());
	Assert.assertEquals(0, summary.getNonNullCount());
	Assert.assertEquals(0, summary.getTrueCount());
	Assert.assertEquals(0, summary.getFalseCount());
}
 
Example #20
Source File: BooleanSummaryAggregatorTest.java    From Flink-CEPplus with Apache License 2.0 5 votes vote down vote up
@Test
public void testMixedGroup() {
	BooleanColumnSummary summary = summarize(true, false, null, true, true, true, false, null, true, false, true);
	Assert.assertEquals(11, summary.getTotalCount());
	Assert.assertEquals(2, summary.getNullCount());
	Assert.assertEquals(9, summary.getNonNullCount());
	Assert.assertEquals(6, summary.getTrueCount());
	Assert.assertEquals(3, summary.getFalseCount());
}
 
Example #21
Source File: BooleanSummaryAggregator.java    From Flink-CEPplus with Apache License 2.0 5 votes vote down vote up
@Override
public void combine(Aggregator<Boolean, BooleanColumnSummary> otherSameType) {
	BooleanSummaryAggregator other = (BooleanSummaryAggregator) otherSameType;
	trueCount += other.trueCount;
	falseCount += other.falseCount;
	nullCount += other.nullCount;
}
 
Example #22
Source File: DataSetUtilsITCase.java    From Flink-CEPplus with Apache License 2.0 4 votes vote down vote up
@Test
public void testSummarize() throws Exception {
	final ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();

	List<Tuple8<Short, Integer, Long, Float, Double, String, Boolean, DoubleValue>> data = new ArrayList<>();
	data.add(new Tuple8<>((short) 1, 1, 100L, 0.1f, 1.012376, "hello", false, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 2, 2, 1000L, 0.2f, 2.003453, "hello", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 4, 10, 10000L, 0.2f, 75.00005, "null", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 10, 4, 100L, 0.9f, 79.5, "", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 5, 5, 1000L, 0.2f, 10.0000001, "a", false, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 6, 6, 10L, 0.1f, 0.0000000000023, "", true, new DoubleValue(100.0)));
	data.add(new Tuple8<>((short) 7, 7, 1L, 0.2f, Double.POSITIVE_INFINITY, "abcdefghijklmnop", true, new DoubleValue(100.0)));
	data.add(new Tuple8<>((short) 8, 8, -100L, 0.001f, Double.NaN, "abcdefghi", true, new DoubleValue(100.0)));

	Collections.shuffle(data);

	DataSet<Tuple8<Short, Integer, Long, Float, Double, String, Boolean, DoubleValue>> ds = env.fromCollection(data);

	// call method under test
	Tuple results = DataSetUtils.summarize(ds);

	Assert.assertEquals(8, results.getArity());

	NumericColumnSummary<Short> col0Summary = results.getField(0);
	Assert.assertEquals(8, col0Summary.getNonMissingCount());
	Assert.assertEquals(1, col0Summary.getMin().shortValue());
	Assert.assertEquals(10, col0Summary.getMax().shortValue());
	Assert.assertEquals(5.375, col0Summary.getMean().doubleValue(), 0.0);

	NumericColumnSummary<Integer> col1Summary = results.getField(1);
	Assert.assertEquals(1, col1Summary.getMin().intValue());
	Assert.assertEquals(10, col1Summary.getMax().intValue());
	Assert.assertEquals(5.375, col1Summary.getMean().doubleValue(), 0.0);

	NumericColumnSummary<Long> col2Summary = results.getField(2);
	Assert.assertEquals(-100L, col2Summary.getMin().longValue());
	Assert.assertEquals(10000L, col2Summary.getMax().longValue());

	NumericColumnSummary<Float> col3Summary = results.getField(3);
	Assert.assertEquals(8, col3Summary.getTotalCount());
	Assert.assertEquals(0.001000, col3Summary.getMin().doubleValue(), 0.0000001);
	Assert.assertEquals(0.89999999, col3Summary.getMax().doubleValue(), 0.0000001);
	Assert.assertEquals(0.2376249988883501, col3Summary.getMean().doubleValue(), 0.000000000001);
	Assert.assertEquals(0.0768965488108089, col3Summary.getVariance().doubleValue(), 0.00000001);
	Assert.assertEquals(0.27730226975415995, col3Summary.getStandardDeviation().doubleValue(), 0.000000000001);

	NumericColumnSummary<Double> col4Summary = results.getField(4);
	Assert.assertEquals(6, col4Summary.getNonMissingCount());
	Assert.assertEquals(2, col4Summary.getMissingCount());
	Assert.assertEquals(0.0000000000023, col4Summary.getMin().doubleValue(), 0.0);
	Assert.assertEquals(79.5, col4Summary.getMax().doubleValue(), 0.000000000001);

	StringColumnSummary col5Summary = results.getField(5);
	Assert.assertEquals(8, col5Summary.getTotalCount());
	Assert.assertEquals(0, col5Summary.getNullCount());
	Assert.assertEquals(8, col5Summary.getNonNullCount());
	Assert.assertEquals(2, col5Summary.getEmptyCount());
	Assert.assertEquals(0, col5Summary.getMinLength().intValue());
	Assert.assertEquals(16, col5Summary.getMaxLength().intValue());
	Assert.assertEquals(5.0, col5Summary.getMeanLength().doubleValue(), 0.0001);

	BooleanColumnSummary col6Summary = results.getField(6);
	Assert.assertEquals(8, col6Summary.getTotalCount());
	Assert.assertEquals(2, col6Summary.getFalseCount());
	Assert.assertEquals(6, col6Summary.getTrueCount());
	Assert.assertEquals(0, col6Summary.getNullCount());

	NumericColumnSummary<Double> col7Summary = results.getField(7);
	Assert.assertEquals(100.0, col7Summary.getMax().doubleValue(), 0.00001);
	Assert.assertEquals(50.0, col7Summary.getMin().doubleValue(), 0.00001);
}
 
Example #23
Source File: BooleanSummaryAggregator.java    From flink with Apache License 2.0 4 votes vote down vote up
@Override
public BooleanColumnSummary result() {
	return new BooleanColumnSummary(trueCount, falseCount, nullCount);
}
 
Example #24
Source File: DataSetUtilsITCase.java    From flink with Apache License 2.0 4 votes vote down vote up
@Test
public void testSummarize() throws Exception {
	final ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();

	List<Tuple8<Short, Integer, Long, Float, Double, String, Boolean, DoubleValue>> data = new ArrayList<>();
	data.add(new Tuple8<>((short) 1, 1, 100L, 0.1f, 1.012376, "hello", false, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 2, 2, 1000L, 0.2f, 2.003453, "hello", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 4, 10, 10000L, 0.2f, 75.00005, "null", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 10, 4, 100L, 0.9f, 79.5, "", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 5, 5, 1000L, 0.2f, 10.0000001, "a", false, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 6, 6, 10L, 0.1f, 0.0000000000023, "", true, new DoubleValue(100.0)));
	data.add(new Tuple8<>((short) 7, 7, 1L, 0.2f, Double.POSITIVE_INFINITY, "abcdefghijklmnop", true, new DoubleValue(100.0)));
	data.add(new Tuple8<>((short) 8, 8, -100L, 0.001f, Double.NaN, "abcdefghi", true, new DoubleValue(100.0)));

	Collections.shuffle(data);

	DataSet<Tuple8<Short, Integer, Long, Float, Double, String, Boolean, DoubleValue>> ds = env.fromCollection(data);

	// call method under test
	Tuple results = DataSetUtils.summarize(ds);

	Assert.assertEquals(8, results.getArity());

	NumericColumnSummary<Short> col0Summary = results.getField(0);
	Assert.assertEquals(8, col0Summary.getNonMissingCount());
	Assert.assertEquals(1, col0Summary.getMin().shortValue());
	Assert.assertEquals(10, col0Summary.getMax().shortValue());
	Assert.assertEquals(5.375, col0Summary.getMean().doubleValue(), 0.0);

	NumericColumnSummary<Integer> col1Summary = results.getField(1);
	Assert.assertEquals(1, col1Summary.getMin().intValue());
	Assert.assertEquals(10, col1Summary.getMax().intValue());
	Assert.assertEquals(5.375, col1Summary.getMean().doubleValue(), 0.0);

	NumericColumnSummary<Long> col2Summary = results.getField(2);
	Assert.assertEquals(-100L, col2Summary.getMin().longValue());
	Assert.assertEquals(10000L, col2Summary.getMax().longValue());

	NumericColumnSummary<Float> col3Summary = results.getField(3);
	Assert.assertEquals(8, col3Summary.getTotalCount());
	Assert.assertEquals(0.001000, col3Summary.getMin().doubleValue(), 0.0000001);
	Assert.assertEquals(0.89999999, col3Summary.getMax().doubleValue(), 0.0000001);
	Assert.assertEquals(0.2376249988883501, col3Summary.getMean().doubleValue(), 0.000000000001);
	Assert.assertEquals(0.0768965488108089, col3Summary.getVariance().doubleValue(), 0.00000001);
	Assert.assertEquals(0.27730226975415995, col3Summary.getStandardDeviation().doubleValue(), 0.000000000001);

	NumericColumnSummary<Double> col4Summary = results.getField(4);
	Assert.assertEquals(6, col4Summary.getNonMissingCount());
	Assert.assertEquals(2, col4Summary.getMissingCount());
	Assert.assertEquals(0.0000000000023, col4Summary.getMin().doubleValue(), 0.0);
	Assert.assertEquals(79.5, col4Summary.getMax().doubleValue(), 0.000000000001);

	StringColumnSummary col5Summary = results.getField(5);
	Assert.assertEquals(8, col5Summary.getTotalCount());
	Assert.assertEquals(0, col5Summary.getNullCount());
	Assert.assertEquals(8, col5Summary.getNonNullCount());
	Assert.assertEquals(2, col5Summary.getEmptyCount());
	Assert.assertEquals(0, col5Summary.getMinLength().intValue());
	Assert.assertEquals(16, col5Summary.getMaxLength().intValue());
	Assert.assertEquals(5.0, col5Summary.getMeanLength().doubleValue(), 0.0001);

	BooleanColumnSummary col6Summary = results.getField(6);
	Assert.assertEquals(8, col6Summary.getTotalCount());
	Assert.assertEquals(2, col6Summary.getFalseCount());
	Assert.assertEquals(6, col6Summary.getTrueCount());
	Assert.assertEquals(0, col6Summary.getNullCount());

	NumericColumnSummary<Double> col7Summary = results.getField(7);
	Assert.assertEquals(100.0, col7Summary.getMax().doubleValue(), 0.00001);
	Assert.assertEquals(50.0, col7Summary.getMin().doubleValue(), 0.00001);
}
 
Example #25
Source File: DataSetUtilsITCase.java    From flink with Apache License 2.0 4 votes vote down vote up
@Test
public void testSummarize() throws Exception {
	final ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();

	List<Tuple8<Short, Integer, Long, Float, Double, String, Boolean, DoubleValue>> data = new ArrayList<>();
	data.add(new Tuple8<>((short) 1, 1, 100L, 0.1f, 1.012376, "hello", false, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 2, 2, 1000L, 0.2f, 2.003453, "hello", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 4, 10, 10000L, 0.2f, 75.00005, "null", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 10, 4, 100L, 0.9f, 79.5, "", true, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 5, 5, 1000L, 0.2f, 10.0000001, "a", false, new DoubleValue(50.0)));
	data.add(new Tuple8<>((short) 6, 6, 10L, 0.1f, 0.0000000000023, "", true, new DoubleValue(100.0)));
	data.add(new Tuple8<>((short) 7, 7, 1L, 0.2f, Double.POSITIVE_INFINITY, "abcdefghijklmnop", true, new DoubleValue(100.0)));
	data.add(new Tuple8<>((short) 8, 8, -100L, 0.001f, Double.NaN, "abcdefghi", true, new DoubleValue(100.0)));

	Collections.shuffle(data);

	DataSet<Tuple8<Short, Integer, Long, Float, Double, String, Boolean, DoubleValue>> ds = env.fromCollection(data);

	// call method under test
	Tuple results = DataSetUtils.summarize(ds);

	Assert.assertEquals(8, results.getArity());

	NumericColumnSummary<Short> col0Summary = results.getField(0);
	Assert.assertEquals(8, col0Summary.getNonMissingCount());
	Assert.assertEquals(1, col0Summary.getMin().shortValue());
	Assert.assertEquals(10, col0Summary.getMax().shortValue());
	Assert.assertEquals(5.375, col0Summary.getMean().doubleValue(), 0.0);

	NumericColumnSummary<Integer> col1Summary = results.getField(1);
	Assert.assertEquals(1, col1Summary.getMin().intValue());
	Assert.assertEquals(10, col1Summary.getMax().intValue());
	Assert.assertEquals(5.375, col1Summary.getMean().doubleValue(), 0.0);

	NumericColumnSummary<Long> col2Summary = results.getField(2);
	Assert.assertEquals(-100L, col2Summary.getMin().longValue());
	Assert.assertEquals(10000L, col2Summary.getMax().longValue());

	NumericColumnSummary<Float> col3Summary = results.getField(3);
	Assert.assertEquals(8, col3Summary.getTotalCount());
	Assert.assertEquals(0.001000, col3Summary.getMin().doubleValue(), 0.0000001);
	Assert.assertEquals(0.89999999, col3Summary.getMax().doubleValue(), 0.0000001);
	Assert.assertEquals(0.2376249988883501, col3Summary.getMean().doubleValue(), 0.000000000001);
	Assert.assertEquals(0.0768965488108089, col3Summary.getVariance().doubleValue(), 0.00000001);
	Assert.assertEquals(0.27730226975415995, col3Summary.getStandardDeviation().doubleValue(), 0.000000000001);

	NumericColumnSummary<Double> col4Summary = results.getField(4);
	Assert.assertEquals(6, col4Summary.getNonMissingCount());
	Assert.assertEquals(2, col4Summary.getMissingCount());
	Assert.assertEquals(0.0000000000023, col4Summary.getMin().doubleValue(), 0.0);
	Assert.assertEquals(79.5, col4Summary.getMax().doubleValue(), 0.000000000001);

	StringColumnSummary col5Summary = results.getField(5);
	Assert.assertEquals(8, col5Summary.getTotalCount());
	Assert.assertEquals(0, col5Summary.getNullCount());
	Assert.assertEquals(8, col5Summary.getNonNullCount());
	Assert.assertEquals(2, col5Summary.getEmptyCount());
	Assert.assertEquals(0, col5Summary.getMinLength().intValue());
	Assert.assertEquals(16, col5Summary.getMaxLength().intValue());
	Assert.assertEquals(5.0, col5Summary.getMeanLength().doubleValue(), 0.0001);

	BooleanColumnSummary col6Summary = results.getField(6);
	Assert.assertEquals(8, col6Summary.getTotalCount());
	Assert.assertEquals(2, col6Summary.getFalseCount());
	Assert.assertEquals(6, col6Summary.getTrueCount());
	Assert.assertEquals(0, col6Summary.getNullCount());

	NumericColumnSummary<Double> col7Summary = results.getField(7);
	Assert.assertEquals(100.0, col7Summary.getMax().doubleValue(), 0.00001);
	Assert.assertEquals(50.0, col7Summary.getMin().doubleValue(), 0.00001);
}
 
Example #26
Source File: BooleanSummaryAggregator.java    From flink with Apache License 2.0 4 votes vote down vote up
@Override
public BooleanColumnSummary result() {
	return new BooleanColumnSummary(trueCount, falseCount, nullCount);
}
 
Example #27
Source File: BooleanSummaryAggregator.java    From Flink-CEPplus with Apache License 2.0 4 votes vote down vote up
@Override
public BooleanColumnSummary result() {
	return new BooleanColumnSummary(trueCount, falseCount, nullCount);
}