version 1.1.0, send to travis-ci

Author: julienvollering

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: MIAmaxent
 Type: Package
 Title: A Modular, Integrated Approach to Maximum Entropy Distribution Modeling
-Version: 1.0.0.9000
+Version: 1.1.0
 Authors@R: c(
     person("Julien", "Vollering", email = "[email protected]", role = c("aut", "cre")),
     person("Sabrina", "Mazzoni", email = "[email protected]", role = "aut"),
@@ -21,6 +21,7 @@ LazyData: TRUE
 URL: https://github.com/julienvollering/MIAmaxent
 BugReports: https://github.com/julienvollering/MIAmaxent/issues
 RoxygenNote: 6.1.1
+Encoding: UTF-8
 Imports:
     dplyr (>= 0.4.3),
     e1071 (>= 1.6-7),

NEWS.md

@@ -1,11 +1,15 @@
-# MIAmaxent 1.0.0.9000
+# MIAmaxent 1.1.0
+
 * Added ellipsis to plotFOP() for easier customization of graphic
+* Feature request: added 'densitythreshold' argument to plotFOP()
+* Predictions NA when a transformation returns NaN. 
 * Minor documentation edits to readData and testAUC
 * Simplification of F-statistic calculation; may cause rounding differences with respect to previous versions.
 
 # MIAmaxent 1.0.0
 
 ## Major changes
+
 * Model fitting implemented as infinitely-weighted logistic regression, so that all computation can be done natively in R (maxent.jar no longer required).
 * Implements choice of algorithm: "maxent" for maximum entropy or "LR" for standard logistic regression (binomial GLM).
 * No files written to system unless write = TRUE

R/MIAmaxent.R

@@ -35,7 +35,7 @@
 #' @references Halvorsen, R., Mazzoni, S., Bryn, A. & Bakkestuen, V. (2015)
 #'   Opportunities for improved distribution modelling practice via a strict
 #'   maximum likelihood interpretation of MaxEnt. Ecography, 38, 172-183.
-#' @references Halvorsen, R., Mazzoni, S., Dirksen, J.W., Naesset, E., Gobakken,
+#' @references Halvorsen, R., Mazzoni, S., Dirksen, J.W., Næsset, E., Gobakken,
 #'   T. & Ohlson, M. (2016) How important are choice of model selection method
 #'   and spatial autocorrelation of presence data for distribution modelling by
 #'   MaxEnt? Ecological Modelling, 328, 108-118.

R/deriveVars.R

@@ -9,8 +9,8 @@
 #'
 #' The linear transformation "L" is a simple rescaling to the range [0, 1].
 #'
-#' The monotonous transformation "M" performed is a zero-skew transformation
-#' (Oekland et al. 2001).
+#' The monotone transformation "M" performed is a zero-skew transformation
+#' (Økland et al. 2001).
 #'
 #' The deviation transformation "D" is performed around an optimum EV value that
 #' is found by looking at frequency of presence (see \code{\link{plotFOP}}).
@@ -49,7 +49,7 @@
 #'   complete (no NAs). See \code{\link{readData}}.
 #' @param transformtype Specifies the types of transformations types to be
 #'   performed. Default is the full set of the following transformation types: L
-#'   (linear), M (monotonous), D (deviation), HF (forward hinge), HR (reverse
+#'   (linear), M (monotone), D (deviation), HF (forward hinge), HR (reverse
 #'   hinge), T (threshold), and B (binary).
 #' @param allsplines Logical. Keep all spline transformations created, rather
 #'   than pre-selecting particular splines based on fraction of total variation
@@ -74,7 +74,7 @@
 #' @references Halvorsen, R., Mazzoni, S., Bryn, A., & Bakkestuen, V. (2015).
 #'   Opportunities for improved distribution modelling practice via a strict
 #'   maximum likelihood interpretation of MaxEnt. Ecography, 38(2), 172-183.
-#' @references Oekland, R.H., Oekland, T. & Rydgren, K. (2001).
+#' @references Økland, R.H., Økland, T. & Rydgren, K. (2001).
 #'   Vegetation-environment relationships of boreal spruce swamp forests in
 #'   Oestmarka Nature Reserve, SE Norway. Sommerfeltia, 29, 1-190.
 #'

R/plotFOP.R

@@ -38,6 +38,9 @@
 #' @param ranging Logical. If \code{TRUE}, will range the EV scale to [0,1].
 #'   This is equivalent to plotting FOP over the linear transformation produced
 #'   by deriveVars. Irrelevant for categorical EVs.
+#' @param densitythreshold Numeric. Intervals containing fewer than this number
+#'   of observations will be represented with an open symbol in the plot.
+#'   Irrelevant for categorical EVs.
 #' @param ... Arguments to be passed to \code{plot} or \code{barplot} to control
 #'   the appearance of the plot. For example: \itemize{ \item \code{lwd} for
 #'   line width \item \code{cex.main} for size of plot title \item \code{space}
@@ -54,10 +57,12 @@
 #'   observations in the level ("n"), and the mean RV value of the level
 #'   ("levelRV") are used.}
 #'
-#' @references Stoea, B., Halvorsen, R., Mazzoni, S. & Gusarov, V. (2016)
+#' @references Støa, B., R. Halvorsen, S. Mazzoni, and V. I. Gusarov. (2018).
 #'   Sampling bias in presence-only data used for species distribution
-#'   modelling: assessment and effects on models. Sommerfeltia, submitted
-#'   manuscript.
+#'   modelling: theory and methods for detecting sample bias and its effects on
+#'   models. Sommerfeltia 38:1–53.
+
+
 #'
 #' @examples
 #' FOPev11 <- plotFOP(toydata_sp1po, 2)
@@ -78,7 +83,8 @@
 #' @export
 
 
-plotFOP <- function(data, EV, span = 0.5, intervals = NULL, ranging = FALSE, ...) {
+plotFOP <- function(data, EV, span = 0.5, intervals = NULL, ranging = FALSE,
+                    densitythreshold = NULL, ...) {
 
   if (EV==1) {
     stop("'EV' cannot be the first column of 'data', which must be the response variable")
@@ -109,8 +115,6 @@ plotFOP <- function(data, EV, span = 0.5, intervals = NULL, ranging = FALSE, ...
     FOP <- list(EVoptimum = evoptimum,
                 FOPdata = FOPdf)
 
-
-
     op <- graphics::par(mar=(c(5, 4, 4, 4) + 0.3))
     on.exit(graphics::par(op))
     dens <- stats::density(df[, 2])
@@ -120,8 +124,11 @@ plotFOP <- function(data, EV, span = 0.5, intervals = NULL, ranging = FALSE, ...
     graphics::axis(side=4, col="grey60", col.axis="grey60", las=1)
     graphics::mtext("Kernel estimated data density", side=4, line=3, col="grey60")
     graphics::par(new=TRUE)
+    if (is.null(densitythreshold)) {densitythreshold <- 0}
+    closedsymbols <- as.numeric(FOPdf$n >= densitythreshold)+1
     args1 <- list(bty="n", main = paste0("FOP plot: ", evname), xlab = evname,
-                  ylab = "Frequency of Observed Presence (FOP)", pch=20, las=1)
+                  ylab = "Frequency of Observed Presence (FOP)", las=1,
+                  pch=c(1,20)[closedsymbols])
     inargs <- list(...)
     args1[names(inargs)] <- inargs
     do.call(graphics::plot, c(list(x=FOPdf$intEV, y=FOPdf$intRV), args1))

R/projectModel.R

@@ -126,11 +126,19 @@ projectModel <- function(model, transformations, data, clamping = FALSE,
   })
   newdata <- as.data.frame(do.call(cbind, dvdatani))
   names(newdata) <- dvnamesni
-
   type <- if (class(model)[1] == "iwlr") {
     ifelse(raw == TRUE, "raw", "PRO")
   } else { "response" }
-  preds <- stats::predict(model, newdata, type)
+
+  if (any(is.nan(unlist(newdata)))) {
+    nans <- which(apply(newdata, 1, function(x) {any(is.nan(x))}))
+    warning("Transformed 'data' has ", length(nans), " rows with NaN. Predictions for these will be NA.",
+            call. = FALSE)
+    preds <- rep(NA, nrow(newdata))
+    preds[-nans] <- stats::predict(model, newdata[-nans,], type)
+      } else {
+    preds <- stats::predict(model, newdata, type)
+  }
 
   if (class(model)[1] == "iwlr" && rescale == TRUE) {
     if (raw == TRUE) { preds <- preds/sum(preds, na.rm = TRUE)

cran-comments.md

@@ -1,9 +1,8 @@
-## Major release submission
+## Minor release submission
 
 ## Test environments
-* local Windows install, R 3.5.1
+* local OS X install, R 3.6.0
 * ubuntu 14.04 (on travis-ci), R 3.5.1
-* os x 10.13, (on travis-ci), R 3.5.0
 * win-builder, R devel
 
 ## R CMD check results