fixed cran comments

Author: gravesee

R/RcppExports.R

@@ -13,7 +13,7 @@ predict_iForest_sparse_nodes <- function(x, Model) {
     .Call('_isofor_predict_iForest_sparse_nodes', PACKAGE = 'isofor', x, Model)
 }
 
-predict_iForest_pathlength_cpp <- function(df, Model, n_cores) {
-    .Call('_isofor_predict_iForest_pathlength_cpp', PACKAGE = 'isofor', df, Model, n_cores)
+predict_iForest_pathlength_cpp <- function(df, Model) {
+    .Call('_isofor_predict_iForest_pathlength_cpp', PACKAGE = 'isofor', df, Model)
 }
 

R/predict.R

@@ -7,8 +7,6 @@
 #'
 #' @param newdata a data.frame to predict
 #' @param ... optional arguments not used.
-#' @param n.cores number of cores to use for prediction of anomaly score. Must
-#' be compiled with openmp. Defaults to 1.
 #' @param nodes if true return nobs x ntrees dim matrix with terminal node ids
 #' @param sparse if true return sparse Matrix of dimension nobs x nTerminalNodes.
 #' Each column represents a terminal node. There are as many ones in each row
@@ -23,14 +21,14 @@
 #' correspond to more isolated observations. If sparse or nodes are set to TRUE,
 #' a matrix of the requested type is returned.
 #' @examples
-#' \donttest{
 #' mod <- iForest(iris, phi=16, nt=5)
 #' score <- predict(mod, newdata = iris)
-#' }
+#' @return A numeric vector of length \code{nrow(newdata)} containing values between zero and one.
+#' Values closer to zero are less likely to be anomalous.
 #' @import Matrix
 #' @importFrom parallel detectCores
 #' @export
-predict.iForest <- function(object, newdata, ..., n.cores=1, nodes = FALSE, sparse = FALSE, replace_missing=TRUE, sentinel=-9999999999) {
+predict.iForest <- function(object, newdata, ..., nodes = FALSE, sparse = FALSE, replace_missing=TRUE, sentinel=-9999999999) {
 
   if (!is.data.frame(newdata)) newdata <- as.data.frame(newdata)
 
@@ -71,7 +69,6 @@ predict.iForest <- function(object, newdata, ..., n.cores=1, nodes = FALSE, spar
   } else if (nodes) {
 	  predict_iForest_nodes_cpp(newdata, object)
   } else {
-    num_cores = as.integer(max(1, min(n.cores, detectCores())))
-    predict_iForest_pathlength_cpp(newdata, object, num_cores)
+    predict_iForest_pathlength_cpp(newdata, object)
   }
 }

README.md

@@ -1,19 +1,5 @@
 [![Build Status](https://travis-ci.org/Zelazny7/isofor.svg?branch=master)](https://travis-ci.org/Zelazny7/isofor)
 
-Installation with openmp on osx
--------------------
-
-```
-brew install llvm
-```
-
-And then edit your ~/.R/Makevars file to look like so:
-
-```
-CXX = /usr/local/opt/llvm/bin/clang
-CXXFLAGS = -I/usr/local/opt/llvm/include -fopenmp
-LDFLAGS = -L/usr/local/opt/llvm/lib -fopenmp=libiomp5
-```
 
 Isolation Forest
 ----------------

cran-comments.md

@@ -15,4 +15,6 @@ There were no ERRORs or WARNINGs or NOTEs.
 * removed dontrun directive in predict method examples
 * added donttest directive per CRAN suggestion -- passes windows now
 * fixed typo in DESCRIPTION - "implemenation -> implementation"
+* added return value description for predict method
+* removed OMP so examples pass on i386
 

man/predict.iForest.Rd

@@ -44,15 +44,14 @@ BEGIN_RCPP
 END_RCPP
 }
 // predict_iForest_pathlength_cpp
-SEXP predict_iForest_pathlength_cpp(SEXP df, List Model, SEXP n_cores);
-RcppExport SEXP _isofor_predict_iForest_pathlength_cpp(SEXP dfSEXP, SEXP ModelSEXP, SEXP n_coresSEXP) {
+SEXP predict_iForest_pathlength_cpp(SEXP df, List Model);
+RcppExport SEXP _isofor_predict_iForest_pathlength_cpp(SEXP dfSEXP, SEXP ModelSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
     Rcpp::traits::input_parameter< SEXP >::type df(dfSEXP);
     Rcpp::traits::input_parameter< List >::type Model(ModelSEXP);
-    Rcpp::traits::input_parameter< SEXP >::type n_cores(n_coresSEXP);
-    rcpp_result_gen = Rcpp::wrap(predict_iForest_pathlength_cpp(df, Model, n_cores));
+    rcpp_result_gen = Rcpp::wrap(predict_iForest_pathlength_cpp(df, Model));
     return rcpp_result_gen;
 END_RCPP
 }
@@ -61,7 +60,7 @@ static const R_CallMethodDef CallEntries[] = {
     {"_isofor_nodes_cpp", (DL_FUNC) &_isofor_nodes_cpp, 4},
     {"_isofor_predict_iForest_nodes_cpp", (DL_FUNC) &_isofor_predict_iForest_nodes_cpp, 2},
     {"_isofor_predict_iForest_sparse_nodes", (DL_FUNC) &_isofor_predict_iForest_sparse_nodes, 2},
-    {"_isofor_predict_iForest_pathlength_cpp", (DL_FUNC) &_isofor_predict_iForest_pathlength_cpp, 3},
+    {"_isofor_predict_iForest_pathlength_cpp", (DL_FUNC) &_isofor_predict_iForest_pathlength_cpp, 2},
     {NULL, NULL, 0}
 };
 

src/RcppExports.cpp

@@ -5,10 +5,6 @@
 #include <math.h>
 #include "predict.h"
 
-#ifdef _OPENMP
-#include <omp.h>
-#endif
-
 using namespace Rcpp;
 
 // Enumeration helpers corresponding to matrix columns or list indices
@@ -123,7 +119,6 @@ void predict_pathlength_cpp(SEXP df, SEXP Tree, double * out, int * current_node
     }
   }  while(!all_terminal);
 
-  #pragma omp critical
   for (int i = 0; i < df_nrows; i++) {
     double size = GET_TREE_ATTR(Tree, Size, current_node[i], nrows);
     out[i] += (depth[i] + cn(size)) / n_trees;
@@ -141,25 +136,22 @@ void predict_pathlength_cpp(SEXP df, SEXP Tree, double * out, int * current_node
  */
 
 // [[Rcpp::export]]
-SEXP predict_iForest_pathlength_cpp(SEXP df, List Model, SEXP n_cores) {
+SEXP predict_iForest_pathlength_cpp(SEXP df, List Model) {
 
   int df_nrows = LENGTH(VECTOR_ELT(df, 0));
   int n_trees = INTEGER(Model[NTREES])[0];
 
   double * pl = (double *) calloc(df_nrows, sizeof pl);
 
-  #pragma omp parallel num_threads(INTEGER(n_cores)[0])
   {
 
     int * current_node = (int *) calloc(df_nrows,  sizeof current_node);
     int * depth = (int *) calloc(df_nrows, sizeof depth);
 
-    #pragma omp for
     for (int i = 0; i < n_trees; i++) {
       predict_pathlength_cpp(df, VECTOR_ELT(Model[FOREST], i), pl, current_node, depth, n_trees);
     }
 
-    #pragma omp critical
     free(current_node);
     free(depth);
   }
@@ -169,7 +161,6 @@ SEXP predict_iForest_pathlength_cpp(SEXP df, List Model, SEXP n_cores) {
 
   double avg = cn(INTEGER(Model[PHI])[0]);
 
-  #pragma omp parallel for
   for( int i = 0; i < df_nrows; i++ ) {
     REAL(res)[i] = pow(2,  -1 * pl[i] / avg);
   }