Detailed Description

template<class T>
class gnsstk::FDiffFilter< T >

This class computes the first difference of the input data array at each point, as well as stats on both the first difference and the data for the Nwind points ending at the current point. The difference, sigmaYX of the difference statistics, and the slope from the data statistics, are stored in an "analysis vector" Avec. The Avec is used to find outliers, gaps and slips in the data; the analysis() method returns a vector of "result" objects FilterHit<T>. The caller should filter and analyze iteratively, since a single call is often not sufficient to find and fix all the problems.

Note: Flags and xdata are required and critical. Outliers must be flagged or removed between iterations, and xdata is needed for gaps and to correct slip magnitudes for the slope of the data. Probably the major weakness of this filter is it tends to find false slips after large gaps.

Definition at line 131 of file FDiffFilter.hpp.

#include <FDiffFilter.hpp>

Classes
class	Analysis

Public Member Functions
int	analysis ()

void	computeRobustSigmaLimit (int &N, T &new_siglim)

bool	doSmallSlips ()
	get the 'small slips' limit More...

bool	doSmallSlips (bool doit)
	set the 'small slips' limit More...

void	dump (std::ostream &s, std::string tag=std::string())

	FDiffFilter (const std::vector< T > &x, const std::vector< T > &d, const std::vector< int > &f)

int	filter (const size_t index=0, int npts=-1)

std::vector< unsigned int >	getHighSigmaIndex ()
	get the indexes of high-sigma points (only if indexHighSigmas(true)) More...

T	getLimit ()
	get the limit More...

unsigned int	getNhighSigma ()

std::vector< FilterHit< T > >	getResults ()
	get results vector of FilterHit More...

T	getSigma ()
	get the sigma More...

unsigned int	getWidth ()
	get the width of the window More...

bool	indexHighSigmas (bool doit)
	set the indexing flag More...

bool	indexingHighSigmas ()
	get the indexing flag More...

void	setLimit (T val)
	set the limit More...

void	setprecision (int p)
	set precision for output More...

void	setSigma (T val)
	set the sigma More...

void	setw (int w)
	set width for output More...

void	setWidth (unsigned int w)
	set the width of the window More...

Private Attributes
std::vector< Analysis >	Avec

const std::vector< T > &	data
	reference to data to be filtered More...

bool	doSmall
	if true, include small slips (<fdlim) in results More...

T	Esiglim
	compute siglim (= robust outlier limit) More...

T	fdlim
	\|first diff\| must be > this to be outlier/slip More...

const std::vector< int > &	flags
	reference to flags, parallel to data, 0 == good More...

int	ilimit
	largest allowed index in data[] is ilimit-1 More...

bool	keepSigIndex
	if true, keep vector of HighSigmaIndex More...

T	madSlope
	robust stats on slope More...

T	medSlope

bool	noflags
	true when flags array is not given More...

bool	noxdata
	true when xdata is not given More...

unsigned int	Nsig
	filter()'s count of high sigma - analysis needed More...

unsigned int	Nwind
	size of sliding window More...

int	osp
	width and precision for dump(), (default 8,3) More...

int	osw

std::vector< FilterHit< T > >	results
	vector of results of analysis() More...

std::vector< unsigned int >	sigIndexes
	saved indexes of Nsig high sigma pts More...

T	siglim
	sigma > this indicates possible outlier/slip More...

const std::vector< T > &	xdata
	reference to xdata - 'time'; if empty use count More...

Friends
class	IterativeFDiffFilter< T >

Constructor & Destructor Documentation

◆ FDiffFilter()

template<class T >

gnsstk::FDiffFilter< T >::FDiffFilter	(	const std::vector< T > &	x,
		const std::vector< T > &	d,
		const std::vector< int > &	f
	)

inline

constructor with two arrays - x is used only in dump(); x and f must exist but may be empty

Parameters

x	const vector<T> of 'times' values,

Note: (x,d,f) all parallel

Parameters

d	const vector<T> of data values
f	const vector<int> of flags, 0 means good. this array may be empty.

Definition at line 205 of file FDiffFilter.hpp.

Member Function Documentation

◆ analysis()

template<class T >

int gnsstk::FDiffFilter< T >::analysis

analyze the output of the filter(), filling the results array with outliers

Returns: the number of outliers

Definition at line 494 of file FDiffFilter.hpp.

◆ computeRobustSigmaLimit()

template<class T >

void gnsstk::FDiffFilter< T >::computeRobustSigmaLimit	(	int &	N,
		T &	new_siglim
	)

after filter(), and before analyze(), compute robust stats on the sigma of first differences, to get a suggested siglim

Parameters

N	int output: computed number of sigmas > new limit, -1 not enough data
new_siglim	T output: computed limit (high outlier) on sigma for analyze

Definition at line 457 of file FDiffFilter.hpp.

◆ doSmallSlips() [1/2]

template<class T >

bool gnsstk::FDiffFilter< T >::doSmallSlips ( )

inline

get the 'small slips' limit

Definition at line 247 of file FDiffFilter.hpp.

◆ doSmallSlips() [2/2]

template<class T >

bool gnsstk::FDiffFilter< T >::doSmallSlips ( bool doit )

inline

set the 'small slips' limit

Definition at line 240 of file FDiffFilter.hpp.

◆ dump()

template<class T >

void gnsstk::FDiffFilter< T >::dump	(	std::ostream &	s,
		std::string	tag = `std::string()`
	)

dump the data and analysis; optionally include a tag at the start of each line, and configure width and precision.

Definition at line 647 of file FDiffFilter.hpp.

◆ filter()

template<class T >

int gnsstk::FDiffFilter< T >::filter	(	const size_t	index = `0`,
		int	npts = `-1`
	)

filter routine that computes the first difference, and uses siglim to mark areas of possible outlier/slips.

Parameters

index	in data arrays at which to start, defaults to 0
npts	number of data to process, defaults to -1, meaning to end of data.

Returns: > 0 number of points in the analysis vector == number good data -1 the data array is too short (< 2) -3 if the flags array is given but shorter than data array This routine clears the analysis vector.

Definition at line 317 of file FDiffFilter.hpp.

◆ getHighSigmaIndex()

template<class T >

std::vector<unsigned int> gnsstk::FDiffFilter< T >::getHighSigmaIndex ( )

inline

get the indexes of high-sigma points (only if indexHighSigmas(true))

Definition at line 273 of file FDiffFilter.hpp.

◆ getLimit()

template<class T >

T gnsstk::FDiffFilter< T >::getLimit ( )

inline

get the limit

Definition at line 231 of file FDiffFilter.hpp.

◆ getNhighSigma()

template<class T >

unsigned int gnsstk::FDiffFilter< T >::getNhighSigma ( )

inline

get the number of high-sigma points : after filter() can tell if possible outliers/slips present

Definition at line 270 of file FDiffFilter.hpp.

◆ getResults()

template<class T >

std::vector<FilterHit<T> > gnsstk::FDiffFilter< T >::getResults ( )

inline

get results vector of FilterHit

Definition at line 266 of file FDiffFilter.hpp.

◆ getSigma()

template<class T >

T gnsstk::FDiffFilter< T >::getSigma ( )

inline

get the sigma

Definition at line 237 of file FDiffFilter.hpp.

◆ getWidth()

template<class T >

unsigned int gnsstk::FDiffFilter< T >::getWidth ( )

inline

get the width of the window

Definition at line 225 of file FDiffFilter.hpp.

◆ indexHighSigmas()

template<class T >

bool gnsstk::FDiffFilter< T >::indexHighSigmas ( bool doit )

inline

set the indexing flag

Definition at line 250 of file FDiffFilter.hpp.

◆ indexingHighSigmas()

template<class T >

bool gnsstk::FDiffFilter< T >::indexingHighSigmas ( )

inline

get the indexing flag

Definition at line 257 of file FDiffFilter.hpp.

◆ setLimit()

template<class T >

void gnsstk::FDiffFilter< T >::setLimit ( T val )

inline

set the limit

Definition at line 228 of file FDiffFilter.hpp.

◆ setprecision()

template<class T >

void gnsstk::FDiffFilter< T >::setprecision ( int p )

inline

set precision for output

Definition at line 263 of file FDiffFilter.hpp.

◆ setSigma()

template<class T >

void gnsstk::FDiffFilter< T >::setSigma ( T val )

inline

set the sigma

Definition at line 234 of file FDiffFilter.hpp.

◆ setw()

template<class T >

void gnsstk::FDiffFilter< T >::setw ( int w )

inline

set width for output

Definition at line 260 of file FDiffFilter.hpp.

◆ setWidth()

template<class T >

void gnsstk::FDiffFilter< T >::setWidth ( unsigned int w )

inline

set the width of the window

Definition at line 222 of file FDiffFilter.hpp.

Friends And Related Function Documentation

◆ IterativeFDiffFilter< T >

template<class T >

friend class IterativeFDiffFilter< T >

friend

Definition at line 134 of file FDiffFilter.hpp.

Member Data Documentation

◆ Avec

template<class T >

std::vector<Analysis> gnsstk::FDiffFilter< T >::Avec

private

vector of Analysis objects, holding first differences and indexes, generated by filter(), used by analyze() and included in dump() output.

Definition at line 165 of file FDiffFilter.hpp.

◆ data

template<class T >

const std::vector<T>& gnsstk::FDiffFilter< T >::data

private

reference to data to be filtered

Definition at line 177 of file FDiffFilter.hpp.

◆ doSmall

template<class T >

bool gnsstk::FDiffFilter< T >::doSmall

private

if true, include small slips (<fdlim) in results

Definition at line 187 of file FDiffFilter.hpp.

◆ Esiglim

template<class T >

T gnsstk::FDiffFilter< T >::Esiglim

private

compute siglim (= robust outlier limit)

Definition at line 184 of file FDiffFilter.hpp.

◆ fdlim

template<class T >

T gnsstk::FDiffFilter< T >::fdlim

private

|first diff| must be > this to be outlier/slip

Definition at line 182 of file FDiffFilter.hpp.

◆ flags

template<class T >

const std::vector<int>& gnsstk::FDiffFilter< T >::flags

private

reference to flags, parallel to data, 0 == good

Definition at line 179 of file FDiffFilter.hpp.

◆ ilimit

template<class T >

int gnsstk::FDiffFilter< T >::ilimit

private

largest allowed index in data[] is ilimit-1

Definition at line 180 of file FDiffFilter.hpp.

◆ keepSigIndex

template<class T >

bool gnsstk::FDiffFilter< T >::keepSigIndex

private

if true, keep vector of HighSigmaIndex

Definition at line 188 of file FDiffFilter.hpp.

◆ madSlope

template<class T >

T gnsstk::FDiffFilter< T >::madSlope

private

robust stats on slope

Definition at line 192 of file FDiffFilter.hpp.

◆ medSlope

template<class T >

T gnsstk::FDiffFilter< T >::medSlope

private

Definition at line 192 of file FDiffFilter.hpp.

◆ noflags

template<class T >

bool gnsstk::FDiffFilter< T >::noflags

private

true when flags array is not given

Definition at line 173 of file FDiffFilter.hpp.

◆ noxdata

template<class T >

bool gnsstk::FDiffFilter< T >::noxdata

private

true when xdata is not given

Definition at line 172 of file FDiffFilter.hpp.

◆ Nsig

template<class T >

unsigned int gnsstk::FDiffFilter< T >::Nsig

private

filter()'s count of high sigma - analysis needed

Definition at line 186 of file FDiffFilter.hpp.

◆ Nwind

template<class T >

unsigned int gnsstk::FDiffFilter< T >::Nwind

private

size of sliding window

Definition at line 185 of file FDiffFilter.hpp.

◆ osp

template<class T >

int gnsstk::FDiffFilter< T >::osp

private

width and precision for dump(), (default 8,3)

Definition at line 171 of file FDiffFilter.hpp.

◆ osw

template<class T >

int gnsstk::FDiffFilter< T >::osw

private

Definition at line 171 of file FDiffFilter.hpp.

◆ results

template<class T >

std::vector<FilterHit<T> > gnsstk::FDiffFilter< T >::results

private

vector of results of analysis()

Definition at line 168 of file FDiffFilter.hpp.

◆ sigIndexes

template<class T >

std::vector<unsigned int> gnsstk::FDiffFilter< T >::sigIndexes

private

saved indexes of Nsig high sigma pts

Definition at line 190 of file FDiffFilter.hpp.

◆ siglim

template<class T >

T gnsstk::FDiffFilter< T >::siglim

private

sigma > this indicates possible outlier/slip

Definition at line 183 of file FDiffFilter.hpp.

◆ xdata

template<class T >

const std::vector<T>& gnsstk::FDiffFilter< T >::xdata

private

reference to xdata - 'time'; if empty use count

Definition at line 176 of file FDiffFilter.hpp.

The documentation for this class was generated from the following file:

FDiffFilter.hpp

Detailed Description

template<class T> class gnsstk::FDiffFilter< T >

Classes

Public Member Functions

Private Attributes

Friends

Constructor & Destructor Documentation

◆ FDiffFilter()

Member Function Documentation

◆ analysis()

◆ computeRobustSigmaLimit()

◆ doSmallSlips() [1/2]

◆ doSmallSlips() [2/2]

◆ dump()

◆ filter()

◆ getHighSigmaIndex()

◆ getLimit()

◆ getNhighSigma()

◆ getResults()

◆ getSigma()

◆ getWidth()

◆ indexHighSigmas()

◆ indexingHighSigmas()

◆ setLimit()

◆ setprecision()

◆ setSigma()

◆ setw()

◆ setWidth()

Friends And Related Function Documentation

◆ IterativeFDiffFilter< T >

Member Data Documentation

◆ Avec

◆ data

◆ doSmall

◆ Esiglim

◆ fdlim

◆ flags

◆ ilimit

◆ keepSigIndex

◆ madSlope

◆ medSlope

◆ noflags

◆ noxdata

◆ Nsig

◆ Nwind

◆ osp

◆ osw

◆ results

◆ sigIndexes

◆ siglim

◆ xdata

template<class T>
class gnsstk::FDiffFilter< T >