com.goldencode.p2j.pattern

Class StatisticsHelper.Library

java.lang.Object

com.goldencode.p2j.pattern.StatisticsHelper.Library

Enclosing class:

StatisticsHelper

public class StatisticsHelper.Library
extends java.lang.Object

Provides a service for calculating statistics. This class is a PatternEngine worker library.

Constructor Summary

Constructors

Constructor and Description
Library()

Method Summary

Modifier and Type	Method and Description
private java.lang.Number	getValueAtRank(java.util.SortedMap rankMap, long rank) Given a sorted map of cumulative numbers of occurrences (ranks) to numeric data values, return the number associated with a specific rank.
java.util.Map	loadDistributions(java.lang.String filename, java.lang.String type) Given an XML filename and a distribution type, load statistics from an XML document and produce a map of all distributions of that type.
java.lang.Object	maximum(java.util.Map distrib) Locate the maximum value of a distribution, where the map's keys are the data values and the map's values are the corresponding numbers of occurrences of these data values in the distribution.
double	mean(java.util.Map distrib) Calculate the weighted average value of the Numbers represented by the keys of the specified map.
double	median(java.util.Map distrib) Calculate the median value of a distribution, where the map's keys are the data values and the map's values are the corresponding numbers of occurrences of these data values in the distribution.
java.lang.Object	minimum(java.util.Map distrib) Locate the minimum value of a distribution, where the map's keys are the data values and the map's values are the corresponding numbers of occurrences of these data values in the distribution.
long	population(java.util.Map distrib) Calculate the total population size of a distribution.
double	standardDeviation(java.util.Map distrib) Calculate the standard deviation of the given distribution using the "sum of the squares" algorithm.

Methods inherited from class java.lang.Object

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

Constructor Detail

Library

public Library()

Method Detail

maximum

public java.lang.Object maximum(java.util.Map distrib)

Locate the maximum value of a distribution, where the map's keys are the data values and the map's values are the corresponding numbers of occurrences of these data values in the distribution. The map's keys must implement Comparable.

Parameters:

distrib - Map of data values to numbers of occurrences. For purposes of this algorithm, the map's values are ignored; only the keys are of interest.

Returns:

The distribution's maximum value, as determined by the natural order of the keys.

minimum

public java.lang.Object minimum(java.util.Map distrib)

Locate the minimum value of a distribution, where the map's keys are the data values and the map's values are the corresponding numbers of occurrences of these data values in the distribution. The map's keys must implement Comparable.

Parameters:

distrib - Map of data values to numbers of occurrences. For purposes of this algorithm, the map's values are ignored; only the keys are of interest.

Returns:

The distribution's minimum value, as determined by the natural order of the keys.

population

public long population(java.util.Map distrib)

Calculate the total population size of a distribution.

Parameters:

distrib - Map of data values to numbers of occurrences (i.e., population for each data value). For purposes of this algorithm, the map's keys are ignored; only the values are of interest.

Returns:

Cumulative population of all distribution buckets.

median

public double median(java.util.Map distrib)

Calculate the median value of a distribution, where the map's keys are the data values and the map's values are the corresponding numbers of occurrences of these data values in the distribution. The map's keys must be instances of Number, since a numeric result is generated.

To locate the median, we sort the values in the distribution in ascending order, then determine which bucket in the distribution contains the middle-ranked value. For distributions with an odd total population, this number represents the median. For an even total population, the two middle-ranked values are averaged to produce the median.

Parameters:

distrib - Map of data values to numbers of occurrences.

Returns:

The median numeric value.

Throws:

java.lang.ClassCastException - if a non-numeric key is encountered in distrib.

mean

public double mean(java.util.Map distrib)

Calculate the weighted average value of the Numbers represented by the keys of the specified map. The average is weighted by the number of occurrences contained in the value of each bucket of the distribution.

Parameters:

distrib - Distribution buckets; mapping of numeric values to the number of occurrences for each. The keys of this map must each be an instance of Number

Returns:

Weighted average of distrib's numeric keys.

Throws:

java.lang.ClassCastException - if a non-numeric key is encountered in distrib.

standardDeviation

public double standardDeviation(java.util.Map distrib)

Calculate the standard deviation of the given distribution using the "sum of the squares" algorithm.

Parameters:

distrib - Distribution buckets; mapping of numeric values to the number of occurrences for each. The keys of this map must each be an instance of Number

Returns:

Standard deviation of the distribution.

Throws:

java.lang.ClassCastException - if a non-numeric key is encountered in distrib.

loadDistributions

public java.util.Map loadDistributions(java.lang.String filename,
                                       java.lang.String type)

Given an XML filename and a distribution type, load statistics from an XML document and produce a map of all distributions of that type. The map's keys are the distribution names; its values are the distributions themselves.

Each distribution is represented as a SortedMap, where each map key represents a "bucket" value, and each map value represents the the number of occurrences of that value in the distribution.

Parameters:

filename - Name of the file containing the XML document to be read.

type - Type of distribution to collect; this corresponds with the "type" attribute of the "distribution" element within the XML document.

Returns:

An unmodifiable map of unmodifiable distribution maps.

Throws:

java.lang.RuntimeException - if any error occurs loading the distribution data from the specified XML file.

getValueAtRank

private java.lang.Number getValueAtRank(java.util.SortedMap rankMap,
                                        long rank)

Given a sorted map of cumulative numbers of occurrences (ranks) to numeric data values, return the number associated with a specific rank. The map's structure defines ranges of ranks, where one key defines the low end of the range, and the next key in an iteration defines the high end of the range.

Parameters:

rankMap - A sorted map whose keys represent the highest rank for a given data value, when a distribution is sorted in ascending order by data value. The map's values represent the data values at each rank.

rank - Target rank.

Returns:

The numeric value associated with the target rank.