Package org.apache.datasketches

Class BinomialBoundsN

java.lang.Object

org.apache.datasketches.BinomialBoundsN

public final class BinomialBoundsN
extends Object

This class enables the estimation of error bounds given a sample set size, the sampling probability theta, the number of standard deviations and a simple noDataSeen flag. This can be used to estimate error bounds for fixed threshold sampling as well as the error bounds calculations for sketches.

Author:

Kevin Lang

Method Summary

Modifier and Type	Method	Description
static double	getLowerBound(long numSamples, double theta, int numSDev, boolean noDataSeen)	Returns the approximate lower bound value
static double	getUpperBound(long numSamples, double theta, int numSDev, boolean noDataSeen)	Returns the approximate upper bound value

Methods inherited from class java.lang.Object

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

Method Details

getLowerBound

public static double getLowerBound(long numSamples, double theta, int numSDev, boolean noDataSeen)

Returns the approximate lower bound value

Parameters:

numSamples - the number of samples in the sample set

theta - the sampling probability

numSDev - the number of "standard deviations" from the mean for the tail bounds. This must be an integer value of 1, 2 or 3.

noDataSeen - this is normally false. However, in the case where you have zero samples and a theta < 1.0, this flag enables the distinction between a virgin case when no actual data has been seen and the case where the estimate may be zero but an upper error bound may still exist.

Returns:

the approximate lower bound value

getUpperBound

public static double getUpperBound(long numSamples, double theta, int numSDev, boolean noDataSeen)

Returns the approximate upper bound value

Parameters:

numSamples - the number of samples in the sample set

theta - the sampling probability

numSDev - the number of "standard deviations" from the mean for the tail bounds. This must be an integer value of 1, 2 or 3.

noDataSeen - this is normally false. However, in the case where you have zero samples and a theta < 1.0, this flag enables the distinction between a virgin case when no actual data has been seen and the case where the estimate may be zero but an upper error bound may still exist.

Returns:

the approximate upper bound value