Main Page | Directories | File List | File Members | Related Pages

Splits.c File Reference

#include "party.h"

Include dependency graph for Splits.c:

Go to the source code of this file.

Functions

void C_split (const double *x, int p, const double *y, int q, const double *weights, int n, const int *orderx, SEXP splitctrl, SEXP linexpcov2sample, SEXP expcovinf, double *cutpoint, double *maxstat, double *statistics)
SEXP R_split (SEXP x, SEXP y, SEXP weights, SEXP orderx, SEXP linexpcov2sample, SEXP expcovinf, SEXP splitctrl)
void C_splitcategorical (const int *codingx, int p, const double *y, int q, const double *weights, int n, double *standstat, SEXP splitctrl, SEXP linexpcov2sample, SEXP expcovinf, double *cutpoint, int *levelset, double *maxstat, double *statistics)
SEXP R_splitcategorical (SEXP x, SEXP codingx, SEXP y, SEXP weights, SEXP linexpcov2sample, SEXP linexpcov, SEXP expcovinf, SEXP splitctrl)


Detailed Description

Binary splits

Author:
Author
hothorn
Date:
Date
2006-08-25 10:53:10 +0200 (Fri, 25 Aug 2006)

Definition in file Splits.c.


Function Documentation

void C_split const double *  x,
int  p,
const double *  y,
int  q,
const double *  weights,
int  n,
const int *  orderx,
SEXP  splitctrl,
SEXP  linexpcov2sample,
SEXP  expcovinf,
double *  cutpoint,
double *  maxstat,
double *  statistics
 

Search for a cutpoint in a ordered variable x maximizing a two-sample statistic w.r.t. (the influence function of ) the response variable y.

Parameters:
x raw numeric measurements
p dimension of the transformation
y values of the influence function
q dimension of the influence function
weights case weights
n number of observations
orderx the ordering of the transformations, i.e. R> order(x)
splitctrl an object of class `SplitControl'
linexpcov2sample an (uninitialized) object of class `LinStatExpectCovar' with p = 1
expcovinf an initialized object of class `ExpectCovarInfluence'
cutpoint return value; pointer to a double for the cutpoint in x
maxstat return value; pointer to a double for the maximal test statistic
statistics return value; pointer to a n-dim double for the statistics

Definition at line 33 of file Splits.c.

References get_minbucket(), get_minprob(), get_tol(), PL2_covarianceSym, PL2_expectationSym, PL2_linearstatisticSym, and PL2_sumweightsSym.

Referenced by C_Node(), C_splitcategorical(), C_surrogates(), and R_split().

Here is the call graph for this function:

void C_splitcategorical const int *  codingx,
int  p,
const double *  y,
int  q,
const double *  weights,
int  n,
double *  standstat,
SEXP  splitctrl,
SEXP  linexpcov2sample,
SEXP  expcovinf,
double *  cutpoint,
int *  levelset,
double *  maxstat,
double *  statistics
 

Search for a cutpoint in a unordered factor x maximizing a two-sample statistic w.r.t. (the influence function of ) the response variable y.

Parameters:
codingx the coding of x, i.e. as.numeric(x)
p dimension of the transformation
y values of the influence function
q dimension of the influence function
weights case weights
n number of observations
codingx the coding of x, i.e. as.numeric(x)
standstat the vector of the standardized statistics for x, y, weights
splitctrl an object of class `SplitControl'
linexpcov2sample an (uninitialized) object of class `LinStatExpectCovar' with p = 1
expcovinf an initialized object of class `ExpectCovarInfluence'
cutpoint return value; pointer to a double for the cutpoint in x
levelset return value; pointer to a p-dim 0/1 integer
maxstat return value; pointer to a double for the maximal test statistic
statistics return value; pointer to a n-dim double for the statistics

Definition at line 217 of file Splits.c.

References C_split(), and get_tol().

Referenced by C_Node(), and R_splitcategorical().

Here is the call graph for this function:

SEXP R_split SEXP  x,
SEXP  y,
SEXP  weights,
SEXP  orderx,
SEXP  linexpcov2sample,
SEXP  expcovinf,
SEXP  splitctrl
 

R-interface to C_split (does not handle ordered y's)

Parameters:
x values of the transformation
y values of the influence function
weights case weights
orderx the ordering of the transformations, i.e. R> order(x)
linexpcov2sample an (uninitialized) object of class `LinStatExpectCovar' with p = 1
expcovinf an initialized object of class `ExpectCovarInfluence'
splitctrl an object of class `SplitControl'

Definition at line 175 of file Splits.c.

References C_split(), ncol(), and nrow().

Here is the call graph for this function:

SEXP R_splitcategorical SEXP  x,
SEXP  codingx,
SEXP  y,
SEXP  weights,
SEXP  linexpcov2sample,
SEXP  linexpcov,
SEXP  expcovinf,
SEXP  splitctrl
 

R-interface to C_splitcategorical (does not handle ordered y's)

Parameters:
x the values of the x-transformation
codingx the coding of x, i.e. as.numeric(x)
y values of the influence function
weights case weights
linexpcov2sample an (uninitialized) object of class `LinStatExpectCovar' with p = 1
linexpcov an initialized object of class `LinStatExpectCovar'
expcovinf an initialized object of class `ExpectCovarInfluence'
splitctrl an object of class `SplitControl'

Definition at line 307 of file Splits.c.

References C_LinStatExpCov(), C_splitcategorical(), C_standardize(), get_dimension(), get_tol(), ncol(), nrow(), PL2_covarianceSym, PL2_expcovinfSym, PL2_expectationSym, and PL2_linearstatisticSym.

Here is the call graph for this function:


Generated on Fri Aug 25 14:31:57 2006 for party by  doxygen 1.4.4