org.jboss.dna.common.statistic
Class DetailedStatistics<T extends Number>

java.lang.Object
  org.jboss.dna.common.statistic.SimpleStatistics<T>
      org.jboss.dna.common.statistic.DetailedStatistics<T>

Type Parameters:

T - the number type for these statistics

@ThreadSafe
public class DetailedStatistics<T extends Number>
extends SimpleStatistics<T>

Encapsulation of the statistics for a series of values to which new values are frequently added. The statistics include the minimum, maximum, total (aggregate sum), mean (average), median, standard deviation and the histogram of the values.

This class uses an efficient running calculation of the mean and standard deviation that is not as susceptible to roundoff errors as other traditional algorithms. The recursive algorithm is as follows, where M is the median value, sigma is the standard deviation, and S is a variable used in the calculation of sigma:

   M(1) = x(1)
   S(1) = 0
   M(k) = M(k-1) + ( x(k) - M(k-1) ) / k
   S(k) = S(k-1) + ( x(k) - M(k-1) ) * (x(k) - M(k))
 

Then, the standard deviation for n values in x is

 sigma = sqrt(S(n) / n)
 

Unlike the other quantities, the median value (the value at which half of the values are greater and half the values are lower) cannot be calculated incrementally. Therefore, this class does record the values so that the median can be properly calculated. This fact should be kept in mind when performing statistics on large numbers of values.

This class is threadsafe.

Constructor Summary

DetailedStatistics(MathOperations<T> operations)

Method Summary

Histogram<T>	getHistogram() Return the histogram of the values.
Histogram<T>	getHistogram(int numSigmas) Return the histogram of the values.
T	getMedian() Return the approximate mean (average) value represented as an instance of the operand type.
double	getMedianValue() Return the median value.
double	getStandardDeviation() Return the standard deviation.
List<T>	getValues() Get the values that have been recorded in these statistics.
String	toString()

Methods inherited from class org.jboss.dna.common.statistic.SimpleStatistics

add, getCount, getMathOperations, getMaximum, getMean, getMeanValue, getMinimum, getTotal, reset

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

DetailedStatistics

public DetailedStatistics(MathOperations<T> operations)

Method Detail

getValues

public List<T> getValues()

Get the values that have been recorded in these statistics. The contents of this list may change if new values are added in another thread.

Returns:

the unmodifiable collection of values, in insertion order

getMedian

public T getMedian()

Return the approximate mean (average) value represented as an instance of the operand type. Note that this may truncate if the operand type is not able to have the required precision. For the accurate mean, see getMedianValue().

Returns:

the mean (average), or 0.0 if the count is 0

getMedianValue

public double getMedianValue()

Return the median value.

Returns:

the median value, or 0.0 if the count is 0

See Also:

getMedian()

getStandardDeviation

public double getStandardDeviation()

Return the standard deviation. The standard deviation is a measure of the variation in a series of values. Values with a lower standard deviation has less variance in the values than a series of values with a higher standard deviation.

Returns:

the standard deviation, or 0.0 if the count is 0 or if all of the values are the same.

getHistogram

public Histogram<T> getHistogram()

Return the histogram of the values. This method returns a histogram where all of the buckets are distributed normally and all have the same width. In this case, the 'numSigmas' should be set to 0. For other variations, see getHistogram(int).

Returns:

the histogram

See Also:

getHistogram(int)

getHistogram

public Histogram<T> getHistogram(int numSigmas)

Return the histogram of the values. This method is capable of creating two kinds of histograms. The first kind is a histogram where all of the buckets are distributed normally and all have the same width. In this case, the 'numSigmas' should be set to 0. See getHistogram().

The second kind of histogram is more useful when most of the data that is clustered near one value. This histogram is focused around the values that are up to 'numSigmas' above and below the median, and all values outside of this range are placed in the first and last bucket.

Parameters:

numSigmas - the number of standard deviations from the median, or 0 if the buckets of the histogram should be evenly distributed

Returns:

the histogram

See Also:

getHistogram()

toString

public String toString()

Overrides:

toString in class SimpleStatistics<T extends Number>