@Test
public void testNextBinomial() {
BinomialDistributionTest testInstance = new BinomialDistributionTest();
int[] densityPoints = testInstance.makeDensityTestPoints();
double[] densityValues = testInstance.makeDensityTestValues();
int sampleSize = 1000;
int length = TestUtils.eliminateZeroMassPoints(densityPoints, densityValues);
BinomialDistribution distribution = (BinomialDistribution) testInstance.makeDistribution();
double[] expectedCounts = new double[length];
long[] observedCounts = new long[length];
for (int i = 0; i < length; i++) {
expectedCounts[i] = sampleSize * densityValues[i];
}
randomData.reSeed(1000);
for (int i = 0; i < sampleSize; i++) {
int value =
randomData.nextBinomial(
distribution.getNumberOfTrials(), distribution.getProbabilityOfSuccess());
for (int j = 0; j < length; j++) {
if (value == densityPoints[j]) {
observedCounts[j]++;
}
}
}
TestUtils.assertChiSquareAccept(densityPoints, expectedCounts, observedCounts, .001);
}
private void testCorrectnessOfErrorFunction(List<Number> inputList) throws Exception {
int inRange = 0;
int numberOfRuns = 1000;
double sampleRatio = 1 / (double) WEIGHT;
double actual = getExpectedValue(inputList);
Random rand = new Random(1);
for (int i = 0; i < numberOfRuns; i++) {
// Compute Sampled Value using sampledList (numberOfRuns times)
ImmutableList.Builder<Number> sampledList = ImmutableList.builder();
for (Number x : inputList) {
if (rand.nextDouble() < sampleRatio) {
sampledList.add(x);
}
}
BlockBuilder builder = getType().createBlockBuilder(new BlockBuilderStatus());
for (Number sample : sampledList.build()) {
if (getType() == BIGINT) {
BIGINT.writeLong(builder, sample.longValue());
} else if (getType() == DOUBLE) {
DOUBLE.writeDouble(builder, sample.doubleValue());
} else {
throw new AssertionError("Can only handle longs and doubles");
}
}
Page page = new Page(builder.build());
page = OperatorAssertion.appendSampleWeight(ImmutableList.of(page), WEIGHT).get(0);
Accumulator accumulator =
getFunction()
.bind(
ImmutableList.of(0),
Optional.<Integer>absent(),
Optional.of(page.getChannelCount() - 1),
getConfidence())
.createAccumulator();
accumulator.addInput(page);
Block result = accumulator.evaluateFinal();
String approxValue =
BlockAssertions.toValues(accumulator.getFinalType(), result).get(0).toString();
double approx = Double.parseDouble(approxValue.split(" ")[0]);
double error = Double.parseDouble(approxValue.split(" ")[2]);
// Check if actual answer lies within [approxAnswer - error, approxAnswer + error]
if (Math.abs(approx - actual) <= error) {
inRange++;
}
}
BinomialDistribution binomial = new BinomialDistribution(numberOfRuns, getConfidence());
int lowerBound = binomial.inverseCumulativeProbability(0.01);
int upperBound = binomial.inverseCumulativeProbability(0.99);
assertTrue(
lowerBound < inRange && inRange < upperBound,
String.format(
"%d out of %d passed. Expected [%d, %d]",
inRange, numberOfRuns, lowerBound, upperBound));
}
/*
* Authored by yiwu on Oct.8.2014
* reference:
* chapter 2.2 of
* http://www.sciencedirect.com/science/article/pii/016794739390115A
*/
private ArrayList<Integer> sample_value_use_binomial() {
BinomialDistribution binom = null;
int cur = 0;
double cdf = 0;
ArrayList<Integer> result = new ArrayList<Integer>(k);
for (int i = 0; i < k - 1; i++) {
if (n == cur) {
result.add(0);
continue;
}
binom = new BinomialDistribution(n - cur, p[i] / (1.0 - cdf));
int x = binom.sample();
cur += x;
cdf += p[i];
result.add(x);
}
result.add(n - cur);
return result;
}
public static void main(String[] args) {
BinomialDistribution b = new BinomialDistribution(5, 0.167);
double p = b.probability(2);
System.out.println(p);
}