#include <Statistic.h>

Inheritance diagram for Statistic< T >:

Collaboration diagram for Statistic< T >:

Public Member Functions
	Statistic ()
	Default constructor.
	Statistic (const T value)
	Constructor from a single value. 'Fakes' a standard deviation.
	Statistic (const Histogram &data, const size_t beginning=0, size_t end=std::numeric_limits< std::size_t >::max())
	Constructor from histogram data.
	Statistic (const Array< T > &data, const size_t beginning=0, size_t end=std::numeric_limits< std::size_t >::max())
	Constructor from array data (NOT histograms)
void	update (const Array< T > &data, const size_t beginning=0, size_t end=std::numeric_limits< std::size_t >::max(), const bool checkForOutliers=false)
	Update an existing Statistic with new data points.
void	update (const Statistic< T > &otherStat)
	Update an existing Statistic with data from an existing statistic.
const double	calcStdev () const
void	update (const T datum)
	Update an existing Statistic with a single new data point.
const bool	similar (const Statistic< T > other, const double alpha=0.05) const
	Uses a two-sided Student T-Test to determine if `other.mean` is 'similar' to the calling object's mean.
const double	tTest (const Statistic< T > other) const
	A two-sided Student T-Test.
const double	nonZeroStdev () const
	Returns stdev if stdev > mean/10, else returns mean/10.
const double	normalisedLikelihood (const double x) const
	The likelihood is normalised by dividing the raw likelihood by the likelihood of the distribution's mean. The end result is that we get a number between 0 and 1 which tells us how close we are to the distribution's mean; if we're right on target then the normalised likelihood will be 1.
void	outputStateBarsLine (std::ostream &o) const
const double	getMean () const
const double	getStdev () const
const T	getMin () const
const T	getMax () const
const size_t	getNumDataPoints () const
Public Attributes
double	mean
double	stdev
T	min
T	max
size_t	numDataPoints
std::list< T >	dataStore
	store every data point we've ever seen so we can re-calculate an accurate stdev on every update.
Friends
std::ostream &	operator<< (std::ostream &o, const Statistic< T > &s)

Detailed Description

template<class T>
struct Statistic< T >

Definition at line 24 of file Statistic.h.

Constructor & Destructor Documentation

template<class T>

Statistic< T >::Statistic ( ) [inline]

Default constructor.

Definition at line 46 of file Statistic.h.

template<class T>

Statistic< T >::Statistic ( const T value ) [inline]

Constructor from a single value. 'Fakes' a standard deviation.

Definition at line 53 of file Statistic.h.

template<class T>

Statistic< T >::Statistic	(	const Histogram &	data,
		const size_t	beginning = `0`,
		size_t	end = `std::numeric_limits<std::size_t>::max()`
	)		`[inline]`

Constructor from histogram data.

Parameters:

data	data
beginning	beginning (gets included in stats)
end	end (excluded from stats)

Definition at line 65 of file Statistic.h.

template<class T>

Statistic< T >::Statistic	(	const Array< T > &	data,
		const size_t	beginning = `0`,
		size_t	end = `std::numeric_limits<std::size_t>::max()`
	)		`[inline]`

Constructor from array data (NOT histograms)

Definition at line 109 of file Statistic.h.

Member Function Documentation

template<class T>

const double Statistic< T >::calcStdev ( ) const [inline]

Definition at line 253 of file Statistic.h.

Referenced by Statistic< Sample_t >::Statistic(), and Statistic< Sample_t >::update().

Here is the caller graph for this function:

template<class T>

const T Statistic< T >::getMax ( ) const [inline]

Definition at line 419 of file Statistic.h.

template<class T>

const double Statistic< T >::getMean ( ) const [inline]

Definition at line 416 of file Statistic.h.

template<class T>

const T Statistic< T >::getMin ( ) const [inline]

Definition at line 418 of file Statistic.h.

template<class T>

const size_t Statistic< T >::getNumDataPoints ( ) const [inline]

Definition at line 420 of file Statistic.h.

template<class T>

const double Statistic< T >::getStdev ( ) const [inline]

Definition at line 417 of file Statistic.h.

template<class T>

const double Statistic< T >::nonZeroStdev ( ) const [inline]

Returns stdev if stdev > mean/10, else returns mean/10.

Definition at line 369 of file Statistic.h.

Referenced by AggregateData::findSpike(), Statistic< Sample_t >::normalisedLikelihood(), and Statistic< Sample_t >::tTest().

Here is the caller graph for this function:

template<class T>

const double Statistic< T >::normalisedLikelihood ( const double x ) const [inline]

The likelihood is normalised by dividing the raw likelihood by the likelihood of the distribution's mean. The end result is that we get a number between 0 and 1 which tells us how close we are to the distribution's mean; if we're right on target then the normalised likelihood will be 1.

Definition at line 383 of file Statistic.h.

Referenced by PowerStateGraph::findBestPath(), and AggregateData::findSpike().

Here is the caller graph for this function:

template<class T>

void Statistic< T >::outputStateBarsLine ( std::ostream & o ) const [inline]

Definition at line 410 of file Statistic.h.

template<class T>

const bool Statistic< T >::similar	(	const Statistic< T >	other,
		const double	alpha = `0.05`
	)		const `[inline]`

Uses a two-sided Student T-Test to determine if other.mean is 'similar' to the calling object's mean.

Returns:: true if other is 'similar' to calling object.

Parameters:

alpha significance level

Definition at line 300 of file Statistic.h.

template<class T>

const double Statistic< T >::tTest ( const Statistic< T > other ) const [inline]

A two-sided Student T-Test.

Uses a Chi-Squared test for equal variances first.

Code adapted from boost tutorial on Comparing the means of two samples with the Students-t test

Returns:: complement of probability (the probability that the difference is due to chance). i.e. the higher this value, the more likely the two distributions have similar means.

Definition at line 325 of file Statistic.h.

Referenced by PowerStateGraph::mostSimilarVertex(), and Statistic< Sample_t >::similar().

Here is the caller graph for this function:

template<class T>

void Statistic< T >::update ( const T datum ) [inline]

Update an existing Statistic with a single new data point.

Update the sample standard deviation with the new data.

Definition at line 270 of file Statistic.h.

template<class T>

void Statistic< T >::update ( const Statistic< T > & otherStat ) [inline]

Update an existing Statistic with data from an existing statistic.

Definition at line 237 of file Statistic.h.

template<class T>

void Statistic< T >::update	(	const Array< T > &	data,
		const size_t	beginning = `0`,
		size_t	end = `std::numeric_limits<std::size_t>::max()`,
		const bool	checkForOutliers = `false`
	)		`[inline]`