version 2.2.2

DESCRIPTION

@@ -1,27 +1,23 @@
 Package: gbm
-Version: 2.1.9
+Version: 2.2.2
 Title: Generalized Boosted Regression Models
 Authors@R: c(
-  person("Ridgeway", "Greg",
+  person("Greg", "Ridgeway",
          email = "gridge@upenn.edu",
          role = c("aut", "cre"),
          comment = c(ORCID = "0000-0001-6911-0804")),
   person("Daniel", "Edwards",
          email = "unknown@unknown.com",
-         role = "ctb",
-         comment = c(ORCID = "0000-0002-8120-0084")),
+         role = "ctb"),
   person("Brian", "Kriegler",
          email = "bkriegler@econone.com",
-         role = "ctb",
-         comment = c(ORCID = "0000-0002-8120-0084")),
+         role = "ctb"),
   person("Stefan", "Schroedl",
          email = "stefan@atomwise.com",
-         role = "ctb",
-         comment = c(ORCID = "0000-0002-8120-0084")),
+         role = "ctb"),
   person("Harry", "Southworth",
          email = "harry@dataclarityconsulting.co.uk",
-         role = "ctb",
-         comment = c(ORCID = "0000-0002-8120-0084")),
+         role = "ctb"),
   person("Brandon", "Greenwell",
          email = "greenwell.brandon@gmail.com",
          role = "ctb",
@@ -52,19 +48,19 @@ License: GPL (>= 2) | file LICENSE
 URL: https://github.com/gbm-developers/gbm
 BugReports: https://github.com/gbm-developers/gbm/issues
 Encoding: UTF-8
-RoxygenNote: 7.2.3
+RoxygenNote: 7.3.1
 VignetteBuilder: knitr
 NeedsCompilation: yes
-Packaged: 2024-01-10 16:15:43 UTC; greg_
-Author: Ridgeway Greg [aut, cre] (<https://orcid.org/0000-0001-6911-0804>),
-  Daniel Edwards [ctb] (<https://orcid.org/0000-0002-8120-0084>),
-  Brian Kriegler [ctb] (<https://orcid.org/0000-0002-8120-0084>),
-  Stefan Schroedl [ctb] (<https://orcid.org/0000-0002-8120-0084>),
-  Harry Southworth [ctb] (<https://orcid.org/0000-0002-8120-0084>),
+Packaged: 2024-06-26 12:33:00 UTC; greg_
+Author: Greg Ridgeway [aut, cre] (<https://orcid.org/0000-0001-6911-0804>),
+  Daniel Edwards [ctb],
+  Brian Kriegler [ctb],
+  Stefan Schroedl [ctb],
+  Harry Southworth [ctb],
   Brandon Greenwell [ctb] (<https://orcid.org/0000-0002-8120-0084>),
   Bradley Boehmke [ctb] (<https://orcid.org/0000-0002-3611-8516>),
   Jay Cunningham [ctb],
   GBM Developers [aut] (https://github.com/gbm-developers)
-Maintainer: Ridgeway Greg <gridge@upenn.edu>
+Maintainer: Greg Ridgeway <gridge@upenn.edu>
 Repository: CRAN
-Date/Publication: 2024-01-10 21:23:15 UTC
+Date/Publication: 2024-06-28 06:20:02 UTC

MD5

@@ -1,18 +1,18 @@
 13d645ea5a474dec3c644512353ee958 *CHANGES
-64b4a2e144395ffcaf8507f483754172 *DESCRIPTION
+e8a38246290de8fe95c46ac1ebc38f2e *DESCRIPTION
 73ed5410da67876bbccbc45fda025419 *LICENSE
 8a3be7d6e9633a3f4a89c866a98df466 *NAMESPACE
 7df0f18225094712a3d207c4486a7db7 *NEWS.md
 30015ca11f94eab451fcfa77bf7283b6 *R/basehaz.gbm.R
 3f6808cb2e1ed79239d9c2b6722399ca *R/calibrate.plot.R
-08efd83992c074fb1c93f5140d80f508 *R/gbm-internals.R
-bd5749be971d74833200f760ade7b286 *R/gbm-package.R
+688bf34a31974b997020f6d3eae12cec *R/gbm-internals.R
+e100ef1e30f39323e8fb5b37f4d8e287 *R/gbm-package.R
 de5d551d0dcd8061a77cbd728bb915b3 *R/gbm.R
-71c3b7b6ab249546493d78a7ac67f3b5 *R/gbm.fit.R
+f2310d9a5225ecacd1a28b13720646b6 *R/gbm.fit.R
 6647c2a0cf4b52f74282d4a30a28c782 *R/gbm.more.R
 0eb9492a4889dd77c7b93ff268e75873 *R/gbm.object.R
 5ddc9dfe12ff465c75b11330bdf4551b *R/gbm.perf.R
-6506a4c195e33ea806a0719e058202a0 *R/gbmCrossVal.R
+2dc03e69665e4599c7289f2993eb3250 *R/gbmCrossVal.R
 b6d4e6c4e10836d532d2d5cf4e8d692b *R/interact.gbm.R
 03f0e1b453cd47ce2da756d9c873627c *R/ir.measures.R
 ca11cea5656713e463f95791318b498f *R/plot.gbm.R
@@ -22,10 +22,10 @@ b3400b409ef852a38f2c93eca941c42d *R/predict.gbm.R
 37f93983b3789b3549b825b052746538 *R/reconstructGBMdata.R
 933e88de5ef87ab875028c5dba6e5829 *R/relative.influence.R
 08f46c17919f160031d6a2f18cd6ef19 *R/test.gbm.R
-a8e091a6102cd9a111ce34609abf0c43 *R/utils.R
+8d5424719baba6f979cc347268487581 *R/utils.R
 b7c53e14f289dc6caa147cb80b381355 *R/zzz.R
 9d96211a993278ccaa571850bf8e4209 *README.md
-7ef4695348ab090cc701fdf459be41d5 *build/vignette.rds
+2a555c7321f4ebc815961fd106c7c578 *build/vignette.rds
 34473f8b2b3d89242031c1af2af38155 *demo/00Index
 c8111e88c09d7a1b3729b41f67b871d1 *demo/OOB-reps.R
 0b6bf4e6c515ffea8a137067f03724d7 *demo/bernoulli.R
@@ -36,15 +36,15 @@ ee94468e32b128e2b7689a61853a998b *demo/coxph.R
 fe8f5f9946ae5cd80cf18c033488d5c5 *demo/printExamples.R
 ac91cb39bbb7b8072ba65cf5933bcf68 *demo/robustReg.R
 94b9c4c04c246fd60085115c9f5a7cf8 *inst/doc/gbm.Rnw
-c1dc0bd6bce467c448752a4b17ed5bb3 *inst/doc/gbm.pdf
+b0ff78e02298d83412611177b001f844 *inst/doc/gbm.pdf
 d7dfb5eccdb0251ad7b9bda805040cdf *inst/tinytest/test_bernoulli.R
 0ecf23a69c73dc027f34b243a9476eae *inst/tinytest/test_coxph.R
 08794176cae150c50cc7bd8e36dbdd23 *inst/tinytest/test_least_squares.R
 0015c0bb9b92caefb3233ca592ef26b8 *inst/tinytest/test_relative_influence.R
 055bd4822b0fa0192e8252df774749f7 *man/basehaz.gbm.Rd
 63c9c494f46f5435ba5929fc6635557c *man/calibrate.plot.Rd
 c2ba2c010f3bf3b1b1d879a9a06fa263 *man/gbm-internals.Rd
-bba5cca188affb5768e0a283a1d9a16b *man/gbm-package.Rd
+710ff7b9b3914708202a6c31c4da448c *man/gbm-package.Rd
 716427ed642502c7a633f6e691a02198 *man/gbm.Rd
 2a033c4b1eb145278f354148c18183b7 *man/gbm.fit.Rd
 b4728cd03e5ca0807ce012514c397747 *man/gbm.more.Rd
@@ -62,54 +62,54 @@ df48c14dc7d0f50f60d3e69eb79ed4fe *man/reconstructGBMdata.Rd
 069434ff43e5f47a2cf584d51d61fd3b *man/relative.influence.Rd
 eeb8e8f08f7d57a76c29169682128fd4 *man/summary.gbm.Rd
 abb04bba52be23fc28a479c5052bcbf2 *man/test.gbm.Rd
-93ccd2f42d70f9e9ad450cbf62b4877c *src/adaboost.cpp
-2f5d22dc3043e69628763cbe303e6b5f *src/adaboost.h
-c877a1d31fa93463ed5d3ccd2164aa80 *src/bernoulli.cpp
-323f73ab809cff64ad5b4f336157f295 *src/bernoulli.h
-088062cab2532d24fa3a9fc5affcf69a *src/buildinfo.h
+9a8e2ff37f4bbc272786d7fb9f018f84 *src/adaboost.cpp
+8bc0e7e6170218dc764317a071ff7929 *src/adaboost.h
+1725d231e272a6353bc694525c5da0f7 *src/bernoulli.cpp
+1b6ea8a45b1a6b9820bf0dbba721acd6 *src/bernoulli.h
+91e43b317576bdf297617babbc600840 *src/buildinfo.h
 e15f767c646f66e54eb5bb20ccd7cebd *src/coxph.cpp
-c8b01f926ddafe1ff39f33f71aed5d0e *src/coxph.h
+f1bea0a96e9bba97336f64aef316659d *src/coxph.h
 3616890b5d7af2b3edd52dc5f29544b0 *src/dataset.cpp
 d30f46362b1915f76e5a328ce95c7136 *src/dataset.h
 b5824ccf353076bf59018429ae3ac6ac *src/distribution.cpp
 91d88e455827695f63bf23df5dfb3108 *src/distribution.h
 6d2bd44a11975c8f023640eb7a9036c3 *src/gaussian.cpp
-6c2bf2616a3b4491aaaf501346246d3f *src/gaussian.h
+ca699b9ef743d6f4c44a469a12dfc7b2 *src/gaussian.h
 27ffff3bcc49d50e130083ef8f2081e5 *src/gbm-init.c
-1d8d4e59887769602b1d3c8dc3d5f94f *src/gbm.cpp
-0f49e8549558916322ec80e29b591a73 *src/gbm.h
+884c1603e6c98b2ae795ed35b07556bd *src/gbm.cpp
+37dfe9bd4a1d1d9e70b9d61c44949b91 *src/gbm.h
 c0c572eb464dae70700ffe8fdc3f6b9f *src/gbm_engine.cpp
-b3f1f49fa614ac6cfd52b28191bfdb70 *src/gbm_engine.h
-c27a350cf827d2decc682ee8ece8332b *src/gbmentry.cpp
+87f344712c06f95ca9ada358aaa3f74a *src/gbm_engine.h
+55715ab97df43bf3f1e62ed2a17e2bb7 *src/gbmentry.cpp
 1fba83f37e9f092d8b005e0c8f32a97b *src/huberized.cpp
-141e5b762944c14a0b6294e15046296f *src/huberized.h
-c5d517b5b38d60df04b9ca6bcc08211a *src/laplace.cpp
-53b4d97c482517fbbc97162da1adf891 *src/laplace.h
-d25bcfb8da3565604f902270b25eb470 *src/locationm.cpp
-932f3d98f158ebf6ae11ed47e873a7f3 *src/locationm.h
-39094967ceaabf7c744bc93d0b86d22f *src/matrix.h
-3cf9f4cf7fe9388b6834b043f587d0e0 *src/multinomial.cpp
-8798fe266a8bad59ac9b3e7019cebbe8 *src/multinomial.h
+e2e3e25937cde555f9099b3099aecf82 *src/huberized.h
+ad0c18190ee958063e3f68564e52b204 *src/laplace.cpp
+a109ba9c2fd1f34dbd04033081580adf *src/laplace.h
+45ede1eab442e5165ecde06bb5e0a028 *src/locationm.cpp
+b79ca7b9c1ece1f8cad0f2958046f436 *src/locationm.h
+5ce6e2bb396ca0b45b97270abae1bb36 *src/matrix.h
+8036edaddcc69c2b8c3f130706f491c7 *src/multinomial.cpp
+2ea82018296bb369e77f96ac9c31c483 *src/multinomial.h
 75737afcbdd3162c62fcdd82b027e1d2 *src/node.cpp
-3f7d35689f88a25a8f536d31c4ce172b *src/node.h
+11d20f2bb5fad44684e8ceef011f490d *src/node.h
 2e99439ebd925c2fe54e609ad93affea *src/node_categorical.cpp
 98afbdcf5bb70211102e58ed262fcec1 *src/node_categorical.h
 a04d827631d157efd47de6ef63633858 *src/node_continuous.cpp
 f09bd89f861430f58cb80ccf0de77c6a *src/node_continuous.h
 af2b9dd107d657344891521829c52243 *src/node_factory.cpp
-3b80b8101a773a42a06eb41b5c6b01c9 *src/node_factory.h
+b39d639f336508ce5b86a4539b754366 *src/node_factory.h
 56dc9a7a6309294654e641c14a32023d *src/node_nonterminal.cpp
 062cbcf913ad61d33048c36ab0b76735 *src/node_nonterminal.h
-a99c0738f82cb857c87b45a65d4e8f25 *src/node_search.cpp
-8cb0947734d82c4bd219ceb6f723c3bb *src/node_search.h
+3f255ece01a44a03fa19e2a1cd4f1c79 *src/node_search.cpp
+4e7e0301f857dae21654e3f1b352a638 *src/node_search.h
 c6943942255ce8138259b6b47caa0c08 *src/node_terminal.cpp
-084bcc63d1b33ca200460b88ef36b8f6 *src/node_terminal.h
-b763976a9c68d9e975417a84b7e2b3c4 *src/pairwise.cpp
-47fc321706b5bcd749bb5b6cc766c36b *src/pairwise.h
+b9753b03cb6ac034a62ad24d7d86c82d *src/node_terminal.h
+67575e1a6a961c63f3898e334bb5d46e *src/pairwise.cpp
+a15faf94a4de6b6e5b7b82834766cc3b *src/pairwise.h
 756422dc1f3f394260fa4d77ec42d1ed *src/poisson.cpp
-0c901877981c1df8c4d82f6dd99c9231 *src/poisson.h
-64e10460138c1b67923020b58cf1a599 *src/quantile.cpp
-491d792d90d047d5a8c192253b632252 *src/quantile.h
+f59ea02f9d947cbba7e37fdba58384b4 *src/poisson.h
+6091d95aba012b027507708aebdaf899 *src/quantile.cpp
+91e37941d9bcb59386adb5b2bc1ea350 *src/quantile.h
 519b30584e7e752480750e86027aea7e *src/tdist.cpp
 9ab15eb81fc9a18ee7d14a76f7aefd2a *src/tdist.h
 276e36bf158250eb458a1cdabcf975b5 *src/tree.cpp

R/gbm-internals.R

@@ -29,7 +29,7 @@
 guessDist <- function(y){
   # If distribution is not given, try to guess it
   if (length(unique(y)) == 2){ d <- "bernoulli" }
-  else if ("Surv" %in% class(y)){ d <- "coxph" }
+  else if (inherits(y, "Surv")){ d <- "coxph" }
   else if (is.factor(y)){ d <- "multinomial" }
   else{ d <- "gaussian" }
   cat(paste("Distribution not specified, assuming", d, "...\n"))

R/gbm-package.R

@@ -25,9 +25,6 @@
 #' 
 #' @useDynLib gbm, .registration = TRUE
 #' 
-#' @name gbm-package
-#' 
-#' @docType package
 #' 
 #' @author Greg Ridgeway \email{gridge@@upenn.edu} with contributions by
 #' Daniel Edwards, Brian Kriegler, Stefan Schroedl, Harry Southworth,
@@ -51,7 +48,9 @@
 #' J.H. Friedman (2002). \dQuote{Stochastic Gradient Boosting,}
 #' \emph{Computational Statistics and Data Analysis} 38(4):367-378.
 #' 
-#' The \url{https://jerryfriedman.su.domains/R-MART.html} website.
+#' The \href{https://jerryfriedman.su.domains/R-MART.html}{MART} website.
 #' 
 #' @keywords package
+"_PACKAGE"
+
 NULL

R/gbm.fit.R

@@ -219,7 +219,7 @@ gbm.fit <- function(x, y, offset = NULL, misc = NULL, distribution = "bernoulli"
   cRows <- nrow(x)
   cCols <- ncol(x)
 
-  if(nrow(x) != ifelse("Surv" %in% class(y), nrow(y), length(y))) {
+  if(nrow(x) != ifelse(inherits(y, "Surv"), nrow(y), length(y))) {
     stop("The number of rows in x does not equal the length of y.")
   }
 
@@ -247,7 +247,7 @@ gbm.fit <- function(x, y, offset = NULL, misc = NULL, distribution = "bernoulli"
   # Check size of data
   if(nTrain * bag.fraction <= 2 * n.minobsinnode + 1) {
     stop("The data set is too small or the subsampling rate is too large: ",
-         "`nTrain * bag.fraction <= n.minobsinnode`")
+         "`nTrain * bag.fraction <= 2 * n.minobsinnode + 1`")
   }
 
   if (distribution$name != "pairwise") {
@@ -314,31 +314,23 @@ gbm.fit <- function(x, y, offset = NULL, misc = NULL, distribution = "bernoulli"
   if(!is.element(distribution$name, supported.distributions)) {
     stop("Distribution ",distribution$name," is not supported")
   }
-  if((distribution$name == "bernoulli") && !all(is.element(y, 0:1)) && 
-     !is.numeric(y)) {
-    # NOTE: Including `!is.numeric(y)` will catch cases where y is a 0/1 factor
-    stop("Bernoulli requires the response to be in {0,1}")
-    if (is.factor(y)) {
-      y <- as.integer(y) - 1
-    }
+  if((distribution$name == "bernoulli") && 
+     (!is.numeric(y) || !all(is.element(y, 0:1)))) {
+    stop("Bernoulli requires the response to be numeric in {0,1}")
   }
-  if((distribution$name == "huberized") && !all(is.element(y,0:1))) {
-    stop("Huberized square hinged loss requires the response to be in {0,1}")
-    if (is.factor(y)) {
-      y <- as.integer(y) - 1
-    }
+  if((distribution$name == "huberized") && 
+     (!is.numeric(y) || !all(is.element(y, 0:1)))) {
+    stop("Huberized square hinged loss requires the response to be numeric in {0,1}")
   }
   if((distribution$name == "poisson") && any(y<0)) {
-    stop("Poisson requires the response to be positive")
+    stop("Poisson requires the response to be non-negative")
   }
   if((distribution$name == "poisson") && any(y != trunc(y))) {
-    stop("Poisson requires the response to be a positive integer")
+    stop("Poisson requires the response to be non-negative")
   }
-  if((distribution$name == "adaboost") && !all(is.element(y,0:1))) {
-    stop("This version of AdaBoost requires the response to be in {0,1}")
-    if (is.factor(y)) {
-      y <- as.integer(y) - 1
-    }
+  if((distribution$name == "adaboost") && 
+     (!is.numeric(y) || !all(is.element(y, 0:1)))) {
+    stop("This version of AdaBoost requires the response to be numeric in {0,1}")
   }
   if(distribution$name == "quantile") {
     if(length(unique(w)) > 1) {
@@ -354,7 +346,7 @@ gbm.fit <- function(x, y, offset = NULL, misc = NULL, distribution = "bernoulli"
     Misc <- c(alpha=distribution$alpha)
   }
   if(distribution$name == "coxph") {
-    if(!("Surv" %in% class(y))) {
+    if(!inherits(y, "Surv")) {
       stop("Outcome must be a survival object Surv(time,failure)")
     }
     if(attr(y,"type")!="right") {

R/gbmCrossVal.R

@@ -133,7 +133,7 @@ gbmCrossValPredictions <- function(cv.models, cv.folds, cv.group,
 
     # The %in% here is to handle coxph
     # my.data  <- data[flag, !(data.names %in% model$response.name)]
-    my.data  <- data[flag, model$var.names]
+    my.data  <- data[flag, model$var.names, drop=FALSE]
     predictions <- predict(model, newdata = my.data, n.trees = best.iter.cv)  # FIXME
     predictions <- matrix(predictions, ncol = num.cols)
     res[flag, ] <- predictions
@@ -204,7 +204,7 @@ gbmDoFold <- function(X, i.train, x, y, offset, distribution, w, var.monotone,
   # Setup
   set.seed(s[[X]])
   i <- order(cv.group == X)
-  x <- x[i.train, , drop = TRUE][i, , drop = FALSE]
+  x <- x[i.train, , drop = FALSE][i, , drop = FALSE]
   y <- y[i.train][i]
   offset <- offset[i.train][i]
   nTrain <- length(which(cv.group != X))

R/utils.R

@@ -56,7 +56,7 @@ best_iter_test <- function(object) {
 best_iter_cv <- function(object) {
   check_if_gbm_fit(object)
   if(!has_cross_validation(object)) {
-    stop('In order to use method="cv" gbm must be called with cv_folds>1.')
+    stop('In order to use method="cv" gbm must be called with cv.folds>1.')
   }
   best_iter_cv <- which.min(object$cv.error)
   return(best_iter_cv)

src/adaboost.cpp

@@ -21,7 +21,7 @@ GBMRESULT CAdaBoost::ComputeWorkingResponse
     double *adWeight,
     bool *afInBag,
     unsigned long nTrain,
-	  int cIdxOff
+    int cIdxOff
 )
 {
     unsigned long i = 0;

src/adaboost.h

@@ -87,8 +87,8 @@ class CAdaBoost : public CDistribution
                           unsigned long nTrain);
 
 private:
-    vector<double> vecdNum;
-    vector<double> vecdDen;
+    std::vector<double> vecdNum;
+    std::vector<double> vecdDen;
 };
 
 #endif // ADABOOST_H

src/bernoulli.cpp

@@ -73,7 +73,7 @@ GBMRESULT CBernoulli::InitF
         double dDen=0.0;         // denominator
         double dNewtonStep=1.0;  // change
         dInitF = 0.0;
-        while(dNewtonStep > 0.0001)
+        while(std::abs(dNewtonStep) > 0.0001)
         {
             dNum=0.0;
             dDen=0.0;

src/bernoulli.h

@@ -88,8 +88,8 @@ class CBernoulli : public CDistribution
                           unsigned long nTrain);
 
 private:
-    vector<double> vecdNum;
-    vector<double> vecdDen;
+    std::vector<double> vecdNum;
+    std::vector<double> vecdDen;
 };
 
 #endif // BERNOULLI_H

src/buildinfo.h

@@ -5,6 +5,7 @@
 #define BUILDINFO_H
 
     #undef ERROR
+    #define R_NO_REMAP // https://rstudio.github.io/r-manuals/r-exts/The-R-API.html
     #include <R.h>
 
     #define GBM_FAILED(hr) ((unsigned long)hr != 0)