FitNgidentity Class Reference

List of all members.

Public Member Functions
	FitNgidentity ()
	FitNgidentity (const int nd)
	~FitNgidentity ()
void	setRuler (const SNavgstd *snv)
int	check (double *ngidens, ostream &log)
double	avgMean () const
double	avgStd () const
void	setCutoff (const double cf)
Static Public Attributes
static double	stdtable [6]
Private Attributes
int	numdb
double *	mean
double *	std
double *	scf
double *	zarray
double *	sczarray
double *	scaledstdarray
double	xsqltcut
double	avgmean
double	avgstd
double	stdslope
double	stdintercept

---

Detailed Description

This class is basically a convient and effient implementation to weed out weak links The private variables are used by the algorithm.

---

Constructor & Destructor Documentation

FitNgidentity::FitNgidentity

(

)

[inline]

FitNgidentity::FitNgidentity

(

const int

nd

)

[inline]

FitNgidentity::~FitNgidentity

(

)

[inline]

References mean, scaledstdarray, scf, sczarray, std, and zarray.

---

Member Function Documentation

void FitNgidentity::setRuler

(

const SNavgstd *

snv

)

given an array of SNavgstd it uses the ngidentity part to initialize this object

References avgmean, avgstd, mean, SNavgstd::meanNgidentity(), numdb, scaledstdarray, scf, sczarray, std, stdintercept, SNavgstd::stdNgidentity(), stdslope, and zarray.

int FitNgidentity::check	(	double *	ngidens,
		ostream &	log
	)

Parameters:

	ngidens	is an array of query to target database ngidentity indexed by the (dbid-1)
	log	is an output stream for log information. This is used for debuging phase only.

Returns:

1 for good, * 0 for gad, * -1 for modified lk: removed some bad link -9 for cannot remove anymore.

remove according to first test

use scaled test results

References avgmean, avgstd, mean, numdb, scaledstdarray, scf, sczarray, std, stdintercept, stdslope, xsqltcut, and zarray.

double FitNgidentity::avgMean

(

)

const [inline]

References avgmean.

double FitNgidentity::avgStd

(

)

const [inline]

References avgstd.

void FitNgidentity::setCutoff

(

const double

cf

)

[inline]

References xsqltcut.

---

Member Data Documentation

double FitNgidentity::stdtable [static]

Initial value:

{ 0.03500686, 0.04219686, 0.05733459,
0.05653033, 0.05211752, 0.04811752 }

int FitNgidentity::numdb [private]

Referenced by check(), readStatInt(), and setRuler().

double* FitNgidentity::mean [private]

the index of the array is the dbid-1 We use 0-based index. mean identity of db1 to all the db2's.

Referenced by check(), setRuler(), and ~FitNgidentity().

double* FitNgidentity::std [private]

array for the std for the whole population

Referenced by check(), setRuler(), and ~FitNgidentity().

double* FitNgidentity::scf [private]

we should use this one for testing. The whole population is too broad. For every 20% range. This is derived by scaling the stdtable. scale factor = std[db2-1]/avgstd

Referenced by check(), setRuler(), and ~FitNgidentity().

double* FitNgidentity::zarray [private]

for efficiency. allocated only once.

Referenced by check(), setRuler(), and ~FitNgidentity().

double* FitNgidentity::sczarray [private]

also for efficiency, stores scaled z-values.

Referenced by check(), setRuler(), and ~FitNgidentity().

double* FitNgidentity::scaledstdarray [private]

for efficiency this should be computed only once.

Referenced by check(), setRuler(), and ~FitNgidentity().

double FitNgidentity::xsqltcut [private]

Referenced by check(), and setCutoff().

double FitNgidentity::avgmean [private]

average over all means

Referenced by avgMean(), check(), and setRuler().

double FitNgidentity::avgstd [private]

Referenced by avgStd(), check(), and setRuler().

double FitNgidentity::stdslope [private]

Referenced by check(), and setRuler().

double FitNgidentity::stdintercept [private]

Referenced by check(), and setRuler().

---

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

• cluster/rmfalselink.cpp