A statistics accumulator. More...

#include <StatAcc.h>

Public Member Functions
	StatAcc ()
	constructors and destructor.
	StatAcc (const StatAcc &)
	~StatAcc ()
void	reset ()
	Reset or copy the accumulator attributes.
void	copy (const StatAcc &)
StatAcc &	operator= (const StatAcc &)
	Operators for adding and copying accumulators.
StatAcc	operator+ (const StatAcc &)
StatAcc &	operator+= (const StatAcc &)
void	put (const T v)
	Accumulate input value(s) v with weight w.
void	put (const T v, const Float w)
void	put (const Array< T > &v)
void	put (const Array< T > &v, const Array< Float > &w)
void	put (const Block< T > &v)
void	put (const Block< T > &v, const Block< Float > &w)
Double	getWtot () const
	Get statistics results one at a time.
uInt	getCount () const
Fallible< Double >	getMin () const
Fallible< Double >	getMax () const
Fallible< Double >	getMean () const
Fallible< Double >	getRms () const
Fallible< Double >	getVariance () const
Fallible< Double >	getRmsAbs () const
void	printSummaryList (std::ostream &, const String &caption) const
	Print summary of accumulated statistics.
void	printSummaryLine (std::ostream &, const String &caption) const
void	printSummaryLineHeader (std::ostream &, const String &caption) const
Private Member Functions
void	put1 (const T, const Float)
	Accumulate a single weighted value.
Private Attributes
Double	itsWtot
Double	itsWsum
Double	itsWssum
Double	itsMin
Double	itsMax
uInt	itsCount

Detailed Description

template<class T>
class casa::StatAcc< T >

A statistics accumulator.

Review Status

Test programs:: tStatAcc

Prerequisite

module Arrays
Arrays module

Etymology

StatAcc stands for `Statistics Accumulator'.

Template Type Argument Requirements (T)

A statistics accumulator accepts (weighted) input values and calculates simple statistice (min, max, weighted mean, rms etc). The accepted input types are real, i.e. Int, uInt, Float, Double, but not Complex. The reason for this is that the < operator of Complex (needed for min/max) works on the norm in any case, and the sqrt function (needed for rms) yields an ambiguous result.

Restriction to real types also allows the internal arithmetic type to be Double rather than the input type. The latter would give all kinds of complications with weighting, accuracy and overflow if the input type would be Int or uInt.

Synopsis

The (weighted) values are fed to StatAcc via the member function `put'. They can be fed individually, or in the form of an Array. The weights are optional (default = 1) and always have type Float.

Asking for a result does not change the internal state. The type of the returned results is always Fallible<Double>. A result is invalid if no input values with non-zero weight have been accumulated yet.

The accumulator can be re-initialised with the function `reset'. Accumulators can be added to each other, which is as if their combined values had been accumulated in the same accumulator.

Some functions have been provided to display a summary of the statistics results. One may choose between a one-line format (with an optional associated header line), and a list.

Example

      StatAcc<T> s;               // T is Float, Double, Int etc
      Matrix<T> vv(2,5);          // a matrix (array) of input values
      Matrix<Float> wgt(2,5);     // an associated matrix of weights
      .... fill vv and wgt with values and individual weights ..\.
      s.put(vv,wgt);              // accumulate the weighted values   
      Fallible<Double> min = s.getMin();    // return the minimum value
   
      s.reset();                  // re-initialise
      s.put(vv);                  // if wgt omitted, default = 1.0
      if (s.getRms().isValid() {  // check validity of rms
            ... use it ..\.
      }

Motivation

One often needs simple statistics of a series of values, which may occur by themselves or in arrays at various points in a program. Sincs it is a pain to have to assign accumulation variables, and to write statistics evaluation code (including exceptions), this helper class is provided.

Definition at line 125 of file StatAcc.h.

Constructor & Destructor Documentation

template<class T>

casa::StatAcc< T >::StatAcc ( )

constructors and destructor.

template<class T>

casa::StatAcc< T >::StatAcc ( const StatAcc< T > & )

template<class T>

casa::StatAcc< T >::~StatAcc ( ) [inline]

Definition at line 131 of file StatAcc.h.

Member Function Documentation

template<class T>

void casa::StatAcc< T >::copy ( const StatAcc< T > & )

template<class T>

uInt casa::StatAcc< T >::getCount ( ) const

template<class T>

Fallible<Double> casa::StatAcc< T >::getMax ( ) const

template<class T>

Fallible<Double> casa::StatAcc< T >::getMean ( ) const

template<class T>

Fallible<Double> casa::StatAcc< T >::getMin ( ) const

template<class T>

Fallible<Double> casa::StatAcc< T >::getRms ( ) const

template<class T>

Fallible<Double> casa::StatAcc< T >::getRmsAbs ( ) const

template<class T>

Fallible<Double> casa::StatAcc< T >::getVariance ( ) const

template<class T>

Double casa::StatAcc< T >::getWtot ( ) const

Get statistics results one at a time.

Count is the nr of values accumulated. Wtot is the sum of the weights. Rms is defined w.r.t. the mean, and is the square of Variance. RmsAbs is the root-mean-square of the absolute input values.

template<class T>

StatAcc casa::StatAcc< T >::operator+ ( const StatAcc< T > & )

template<class T>

StatAcc& casa::StatAcc< T >::operator+= ( const StatAcc< T > & )

template<class T>

StatAcc& casa::StatAcc< T >::operator= ( const StatAcc< T > & )

Operators for adding and copying accumulators.

template<class T>

void casa::StatAcc< T >::printSummaryLine	(	std::ostream &	,
		const String &	caption
	)			const

template<class T>

void casa::StatAcc< T >::printSummaryLineHeader	(	std::ostream &	,
		const String &	caption
	)			const

template<class T>

void casa::StatAcc< T >::printSummaryList	(	std::ostream &	,
		const String &	caption
	)			const

Print summary of accumulated statistics.

Line is a one-line summary, including the (short) caption. LineHeader gives a one-line explanation of the numbers. List uses a separate line for each result (mean, max etc).

template<class T>