org.apache.spark.util.random

Class BinomialBounds

java.lang.Object

org.apache.spark.util.random.BinomialBounds

public class BinomialBounds
extends java.lang.Object

Utility functions that help us determine bounds on adjusted sampling rate to guarantee exact sample size with high confidence when sampling without replacement.

Constructor Summary

Constructors

Constructor and Description
BinomialBounds()

Method Summary

Methods

Modifier and Type	Method and Description
static double	getLowerBound(double delta, long n, double fraction) Returns a threshold p such that if we conduct n Bernoulli trials with success rate = p, it is very unlikely to have more than fraction * n successes.
static double	getUpperBound(double delta, long n, double fraction) Returns a threshold p such that if we conduct n Bernoulli trials with success rate = p, it is very unlikely to have less than fraction * n successes.
static double	minSamplingRate()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

BinomialBounds

public BinomialBounds()

Method Detail

minSamplingRate

public static double minSamplingRate()

getLowerBound

public static double getLowerBound(double delta,
                   long n,
                   double fraction)

Returns a threshold p such that if we conduct n Bernoulli trials with success rate = p, it is very unlikely to have more than fraction * n successes.

Parameters:

delta - (undocumented)

n - (undocumented)

fraction - (undocumented)

Returns:

(undocumented)

getUpperBound

public static double getUpperBound(double delta,
                   long n,
                   double fraction)

Returns a threshold p such that if we conduct n Bernoulli trials with success rate = p, it is very unlikely to have less than fraction * n successes.

Parameters:

delta - (undocumented)

n - (undocumented)

fraction - (undocumented)

Returns:

(undocumented)