how ctree is created in R package 'party'

Question

I want to get the part of R code in package 'party' which creates a Conditional Inference tree. Namely, the part of code which specifies way to grow a tree step by step with a bootstrap sample. I examined the function 'ctree', but it seems to be complicatedly structured. Even I find no sign of such as sample(mtry features from all features). Anyone know where to find it? Or the method how to find it?

Answer 1

It's a C function. Download the package source from http://cran.r-project.org/src/contrib/party_1.0-13.tar.gz and look in src/TreeGrow.c:

/**
    The tree growing recursion
    *\file TreeGrow.c
    *\author $Author: hothorn $
    *\date $Date: 2009-06-16 09:17:31 +0200 (Tue, 16 Jun 2009) $
*/

#include "party.h"


/**
    The main tree growing function, handles the recursion. \n
    *\param node  a list representing the current node
    *\param learnsample an object of class `LearningSample'
    *\param fitmem an object of class `TreeFitMemory'
    *\param controls an object of class `TreeControl'
    *\param where a pointer to an integer vector of n-elements
    *\param nodenum a pointer to a integer vector of length 1
    *\param depth an integer giving the depth of the current node
*/

void C_TreeGrow(SEXP node, SEXP learnsample, SEXP fitmem,
                SEXP controls, int *where, int *nodenum, int depth) {

    SEXP weights;
    int nobs, i, stop;
    double *dweights;

    weights = S3get_nodeweights(node);

    /* stop if either stumps have been requested or
       the maximum depth is exceeded */
    stop = (nodenum[0] == 2 || nodenum[0] == 3) &&
           get_stump(get_tgctrl(controls));
    stop = stop || !check_depth(get_tgctrl(controls), depth);

    if (stop)
        C_Node(node, learnsample, weights, fitmem, controls, 1, depth);
    else
        C_Node(node, learnsample, weights, fitmem, controls, 0, depth);

    S3set_nodeID(node, nodenum[0]);

    if (!S3get_nodeterminal(node)) {

        C_splitnode(node, learnsample, controls);

        /* determine surrogate splits and split missing values */
        if (get_maxsurrogate(get_splitctrl(controls)) > 0) {
            C_surrogates(node, learnsample, weights, controls, fitmem);
            C_splitsurrogate(node, learnsample);
        }

        nodenum[0] += 1;
        C_TreeGrow(S3get_leftnode(node), learnsample, fitmem,
                   controls, where, nodenum, depth + 1);

        nodenum[0] += 1;
        C_TreeGrow(S3get_rightnode(node), learnsample, fitmem,
                   controls, where, nodenum, depth + 1);

    } else {
        dweights = REAL(weights);
        nobs = get_nobs(learnsample);
        for (i = 0; i < nobs; i++)
            if (dweights[i] > 0) where[i] = nodenum[0];
    }
}


/**
    R-interface to C_TreeGrow\n
    *\param learnsample an object of class `LearningSample'
    *\param weights a vector of case weights
    *\param fitmem an object of class `TreeFitMemory'
    *\param controls an object of class `TreeControl'
    *\param where a vector of node indices for each observation
*/

SEXP R_TreeGrow(SEXP learnsample, SEXP weights, SEXP fitmem, SEXP controls, SEXP where) {

     SEXP ans, nweights;
     double *dnweights, *dweights;
     int nobs, i, nodenum = 1;

     GetRNGstate();

     nobs = get_nobs(learnsample);
     PROTECT(ans = allocVector(VECSXP, NODE_LENGTH));
     C_init_node(ans, nobs, get_ninputs(learnsample), get_maxsurrogate(get_splitctrl(controls)),
                 ncol(get_predict_trafo(GET_SLOT(learnsample, PL2_responsesSym))));

     nweights = S3get_nodeweights(ans);
     dnweights = REAL(nweights);
     dweights = REAL(weights);
     for (i = 0; i < nobs; i++) dnweights[i] = dweights[i];

     C_TreeGrow(ans, learnsample, fitmem, controls, INTEGER(where), &nodenum, 1);

     PutRNGstate();

     UNPROTECT(1);
     return(ans);
}