The BNRC score function. More...

#include <BNRC.h>

Public Member Functions
void *	BNRC_init (ytData data, ytKeyValues args)
	Initializes the BNRC score.

void	BNRC_reinit (void buff, ytData data)
	Re-initializes the score function.

double	BNRC_score (void buff, const int j, const int parents, const int q)
	BNRC score function.

void	BNRC_Data_finalize (BNRC_Data *D, int allocXY)
	Releases memory allocated by BNRC_Data_alloc().

size_t	BNRC_Work_alloc (BNRC_Work *W, int p, int n, int mp, int H, int M, int level, int T1_size, int dry)
	Allocates memory for BNRC_Work.

void	BNRC_prepare_beta (double hyper_bg, double hyper_inc, int hyper_num, double beta, double log_beta, int verbose)
	Prepares the hyperparamters.

void	BNRC_calc_Bj (BNRC_Data D, int ofs, int j, int bj, int n, int M, double T1)
	Calculates the design matrix of B-splines.

void	BNRC_calc_B (BNRC_Data D, int ofs, const char type, int n, int M, BNRC_Work *W)
	Calculates B-spline design matrices for all the variables.

void	BNRC_calc_Lambda_j (BNRC_Data D, int ofs, int j, int n, int M, const double beta, int H, double K, BNRC_Work W, double logDetLambda, double LambdaInv, double *LambdaInvBt)
	Calculates Λ related values for the j-th variable.

void	BNRC_calc_Lambda (BNRC_Data D, BNRC_Global G, int p, int n, const char type, BNRC_Work W, int level)
	Calculates Λ related values for all the variables.

void	BNRC_prepare_Lambda (BNRC_Data D, int ofs, const int parents, int q, int n, int level, BNRC_Work W, BNRC_Global G)
	Prepares Lambda.

double	BNRC_calc (BNRC_Data D, int ofs, int j, const int parents, int q, int n, int nn, int level, BNRC_Work W, BNRC_Model P, BNRC_Global *G)
	Calculates BNRC score.

void	BNRC_setEdgeProp (void buff, const int j, const int parents, const int q, const int k, const ytEdge *edge)
	Sets parameters from ytEdge.

void	BNRC_range (BNRC_Data D, const ytDoubleArray xlar, const ytDoubleArray xrar, const char type, double outer)
	Sets the value range xl and xr.

Detailed Description

The BNRC score function.

The BNRC score function accepts only the continuous variables. Missing values are acceptable. The BNRC score function supports changing the number of B-splines for non-parametric regression. Use "M" arguments when initializing the score function.

See also: BNDC

Member Function Documentation

◆ BNRC_calc_B()

void BNRC_calc_B	(	BNRC_Data *	D,
		int	ofs,
		const char *	type,
		int	n,
		int	M,
		BNRC_Work *	W
	)

Calculates B-spline design matrices for all the variables.

This does not perform the calculation for type[j] == 1.

This is thread-enabled.

◆ BNRC_calc_Bj()

void BNRC_calc_Bj	(	BNRC_Data *	D,
		int	ofs,
		int	j,
		int	bj,
		int	n,
		int	M,
		double *	T1
	)

Calculates the design matrix of B-splines.

Calculates the design matrix for the j -th variable (column) of D->X of n samples. The starting position of the input matrix can be offset by specifying ofs.

The calculated design matrix has 4 × n non-zero values and is stored at bj -th position of D->B and D->Bindex both offset by ofs.

The missing values specified by D->Xm are converted to zero values so that these affect nothing in modeling. This also replaces missing values in D->X with zero values.

Parameters

ofs	Offset position for D->X, D->Xm, D->B, and D->Bindex.

◆ BNRC_calc_Lambda()

void BNRC_calc_Lambda	(	BNRC_Data *	D,
		BNRC_Global *	G,
		int	p,
		int	n,
		const char *	type,
		BNRC_Work *	W,
		int	level
	)

Calculates Λ related values for all the variables.

Λ = B^T B + n β K

This is applicable to only the calculation in BNRC_init() and BNRC_reinit() for level >= 2.

◆ BNRC_calc_Lambda_j()

void BNRC_calc_Lambda_j	(	BNRC_Data *	D,
		int	ofs,
		int	j,
		int	n,
		int	M,
		const double *	beta,
		int	H,
		double *	K,
		BNRC_Work *	W,
		double *	logDetLambda,
		double *	LambdaInv,
		double *	LambdaInvBt
	)

Calculates Λ related values for the j-th variable.

This calculates Λ = B^TB + n β K, |Λ^-1| (determinant of Λ), Λ^-1, and Λ^-1 B^T.

Parameters

LambdaInv This can be NULL. This is to store Λ^T for level = 2 calculation.

◆ BNRC_Data_finalize()

void BNRC_Data_finalize	(	BNRC_Data *	D,
		int	allocXY
	)

Releases memory allocated by BNRC_Data_alloc().

Parameters

allocXY Set the same value used when calling BNDC_Data_alloc().

◆ BNRC_init()

void * BNRC_init	(	ytData *	data,
		ytKeyValues *	args
	)

Initializes the BNRC score.

M=n

The number of B-splines for nonparametric regression. By default, n=20.

max_parents=n

mp=n

Maximum parents

prop=n

Property output type. This changes properties to be stored in edges and nodes after the estimation.
0: standard.
1: less information (simple) for the contitous or discrete only variable data set.

hyper_bg=x

hb=x

hb value for the hyperparameter range. The hyperparameter β is determined by the grid search. The grid search starts from β = 10^hb and then decreases the value such as β = 10^{hb − (i × hi )}, where i = 1, 2, ...., hn. (default: x = 2.0)

hyper_inc=x

hi=x

hi value for hyperparameter range. See explanation of hyper_bg. (default: x = 0.4)

hyper_n=n

hn=n

hn value for hyperparameter range. See explanation of hyper_bg. (default: n = 21)

linear

Linear mode. This is short for "hb=2.0,hi=1.0,hn=2".

max_loops=n

The maximum number of loops for parameter estimation by the back fitting algorithm. (default: n = 100)

stop

If specified, the algorithm stops if the parameter estimation is not converged.

verbose=n

v=n

Verbose level. By defalt, n=0.

The following keys are used internally.

xl: DoubleArray instance that defines the minimum (left-most) values of the value ranges for modeling with B-splines.
xr: DoubleArray instance that defines the maximum (right-most) values of the value ranges for modeling with B-splines.

Parameters

data
args

Returns: Pointer to the score buffer.

◆ BNRC_prepare_beta()

void BNRC_prepare_beta	(	double	hyper_bg,
		double	hyper_inc,
		int	hyper_num,
		double *	beta,
		double *	log_beta,
		int	verbose
	)

Prepares the hyperparamters.

Parameters

beta	hyper_num length array.
log_beta	hyper_num length array.

◆ BNRC_prepare_Lambda()

void BNRC_prepare_Lambda	(	BNRC_Data *	D,
		int	ofs,
		const int *	parents,
		int	q,
		int	n,
		int	level,
		BNRC_Work *	W,
		BNRC_Global *	G
	)

Prepares Lambda.

Parameters

ofs	offset. This is ignored for level ≤ 2 calculation.

◆ BNRC_range()

void BNRC_range	(	BNRC_Data *	D,
		const ytDoubleArray *	xlar,
		const ytDoubleArray *	xrar,
		const char *	type,
		double	outer
	)

Sets the value range xl and xr.

Finds the min and max values for every variable. Sets these values in D->xl[j] and D->xr[j] for the j-th variable. These are used in calculating the B-spline design matrix.

If xlar and xrar values are set, this uses them. Otherwise, this finds the min and max values from the data.

Parameters

xlar
xrar

◆ BNRC_reinit()

void BNRC_reinit	(	void *	buff,
		ytData *	data
	)

Re-initializes the score function.

This is only applicable if only the values in data are changed.

Parameters

buff	pointer to the buffere returned by BNRC_init().
data	Only ytData::X and ytData::Y are used.

◆ BNRC_score()

double BNRC_score	(	void *	buff,
		const int	j,
		const int *	parents,
		const int	q
	)

BNRC score function.

This is thread-safe with OpenMP multi-threading. Thread-dependent working memory is allocated in buff returned by BNRC_init(). Therefore, you do not need to initialize and obtain the score buffer for different threads.

Parameters

buff	score buffer returned by BNRC_init().
j	target node.
parents	parent indices.
q	length of parents.

Returns: BNRC score

◆ BNRC_setEdgeProp()

void BNRC_setEdgeProp	(	void *	buff,
		const int	j,
		const int *	parents,
		const int	q,
		const int	k,
		const ytEdge *	edge
	)

Sets parameters from ytEdge.

Sets coefficients (gamma) from "bspline" properties, and calculates B gamma (B_gamma) where B represents the design matrix.

By calling this, BNRC_partialResidual() can calculates valid partial residuals without the score calculation.

◆ BNRC_Work_alloc()

size_t BNRC_Work_alloc	(	BNRC_Work *	W,
		int	p,
		int	n,
		int	mp,
		int	H,
		int	M,
		int	level,
		int	T1_size,
		int	dry
	)

Allocates memory for BNRC_Work.

Parameters

work	whether this allocates BNRC_Work::logDetLambdaWork and BNRC_Work::LambdaInvBtWork.

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

score/BNRC.c

Public Member Functions

Detailed Description

Member Function Documentation

◆ BNRC_calc_B()

◆ BNRC_calc_Bj()

◆ BNRC_calc_Lambda()

◆ BNRC_calc_Lambda_j()

◆ BNRC_Data_finalize()

◆ BNRC_init()

◆ BNRC_prepare_beta()

◆ BNRC_prepare_Lambda()

◆ BNRC_range()

◆ BNRC_reinit()

◆ BNRC_score()

◆ BNRC_setEdgeProp()

◆ BNRC_Work_alloc()