skpr v1.6.3: Fix augmented designs not matching factor and character …

…columns from candidate set

-Add `plot` argument to `plot_correlations()`
-Ensure errors are removed from parallel output

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: skpr
 Title: Design of Experiments Suite: Generate and Evaluate Optimal Designs
-Date: 2024-02-06
-Version: 1.6.2
+Date: 2024-02-20
+Version: 1.6.3
 Authors@R: c(person("Tyler", "Morgan-Wall", email = "tylermw@gmail.com", role = c("aut", "cre")),
              person("George", "Khoury", email = "george.m.khoury@gmail.com", role = c("aut")))
 Description: Generates and evaluates D, I, A, Alias, E, T, and G optimal designs. Supports generation and evaluation of blocked and split/split-split/.../N-split plot designs. Includes parametric and Monte Carlo power evaluation functions, and supports calculating power for censored responses. Provides a framework to evaluate power using functions provided in other packages or written by the user. Includes a Shiny graphical user interface that displays the underlying code used to create and evaluate the design to improve ease-of-use and make analyses more reproducible. For details, see Morgan-Wall et al. (2021) <doi:10.18637/jss.v099.i01>.

R/eval_design_custom_mc.R

@@ -284,6 +284,7 @@ eval_design_custom_mc = function(design, model = NULL, alpha = 0.05,
     run_search = function(iterations) {
        prog = progressr::progressor(steps = nsim)
        foreach::foreach(i = iterations,
+                        .errorhandling = "remove",
                         .options.future = list(packages = parallel_pkgs,
                                                globals = c("extractPvalues", "pvalfunction",
                                                            "parameter_names", "progress", "progressbarupdates",

R/eval_design_mc.R

@@ -753,6 +753,7 @@ eval_design_mc = function(design, model = NULL, alpha = 0.05,
     run_search = function(iterations, is_shiny) {
       prog = progressr::progressor(steps = nsim)
       foreach::foreach (j = seq_along(iterations), .combine = "rbind",
+                        .errorhandling = "remove",
                         .options.future = list(packages = packagelist,
                                                globals = c("extractPvalues", "effectpowermc", "RunMatrixReduced", "is_shiny", "blocking",
                                                            "responses", "contrastslist", "model_formula", "glmfamily", "glmfamilyname", "calceffect",

R/eval_design_survival_mc.R

@@ -328,6 +328,7 @@ eval_design_survival_mc = function(design, model = NULL, alpha = 0.05,
     run_search = function(iterations, is_shiny, surv_args) {
       prog = progressr::progressor(steps = nsim)
       foreach::foreach(i = iterations,
+                       .errorhandling = "remove",
                        .options.future = list(packages = "survival",
                                               globals  = c("extractPvalues", "rfunctionsurv", "parameter_names", "progress", "progressbarupdates",
                                                             "model", "distribution", "RunMatrixReduced", "ModelMatrix", "anticoef" ,"nc", "prog",

R/gen_design.R

@@ -574,7 +574,15 @@ gen_design = function(candidateset, model, trials,
               "' don't match candidate set '",paste0(cand_coltype[aug_coltype != cand_coltype],collapse=", "),
               "'--attempting to covert columns to that of the candidate set")
       for(i in seq_len(ncol(augmentdesign))) {
-        augmentdesign[,i] = methods::as(augmentdesign[,i],cand_coltype[i])
+        if(cand_coltype[i] == "factor") {
+          candsetlevels = levels(candidateset[,i])
+          augmentlevels = unique(augmentdesign[,i, drop = TRUE])
+          augmentlevels_unique = augmentlevels[!augmentlevels %in% candsetlevels]
+          candsetlevels_all = c(candsetlevels,augmentlevels_unique)
+          augmentdesign[,i] = factor(augmentdesign[,i, drop = TRUE],levels=candsetlevels_all)
+        } else {
+          augmentdesign[,i] = methods::as(augmentdesign[,i, drop = TRUE],cand_coltype[i])
+        }
       }
     }
     if (nrow(augmentdesign) >= trials) {
@@ -944,6 +952,7 @@ gen_design = function(candidateset, model, trials,
       run_search = function(iterations, is_shiny) {
         prog = progressr::progressor(steps = repeats)
         foreach::foreach(i = iterations,
+                         .errorhandling = "remove",
                          .options.future = list(globals = c("genOptimalDesign","genBlockedOptimalDesign", "candidatesetmm", "trials",
                                                             "initialreplace", "augmentdesign", "augmentedrows", "augmentdesignmm",
                                                             "progress", "is_shiny", "progressbarupdates", "repeats", "num_updates",
@@ -1059,13 +1068,15 @@ gen_design = function(candidateset, model, trials,
       nc = future::nbrOfWorkers()
       run_search = function(iterations, is_shiny) {
         prog = progressr::progressor(steps = repeats)
-        foreach::foreach(i = iterations, .options.future = list(globals = c("genSplitPlotOptimalDesign", "candidatesetmm", "trials", "candidateset",
-                                                                            "initialreplace", "blockedmm", "interactionlist", "disallowedcomb",
-                                                                            "anydisallowed",
-                                                                            "progress", "is_shiny", "progressbarupdates", "repeats", "num_updates",
-                                                                            "initialdesign", "optimality", "mm", "aliasmm", "blockedmodelmatrix",
-                                                                            "minDopt", "tolerance",  "kexchange" ,"V" ,"prog" ,"nc"),
-                                                                seed = TRUE)) %dofuture% {
+        foreach::foreach(i = iterations,
+                         .errorhandling = "remove",
+                         .options.future = list(globals = c("genSplitPlotOptimalDesign", "candidatesetmm", "trials", "candidateset",
+                                                            "initialreplace", "blockedmm", "interactionlist", "disallowedcomb",
+                                                            "anydisallowed",
+                                                            "progress", "is_shiny", "progressbarupdates", "repeats", "num_updates",
+                                                            "initialdesign", "optimality", "mm", "aliasmm", "blockedmodelmatrix",
+                                                            "minDopt", "tolerance",  "kexchange" ,"V" ,"prog" ,"nc"),
+                                                seed = TRUE)) %dofuture% {
           if(progress || is_shiny) {
             if(is_shiny && i %in% progressbarupdates) {
               prog(sprintf(" (%i workers) ", nc), amount = repeats/num_updates)
@@ -1092,7 +1103,7 @@ gen_design = function(candidateset, model, trials,
   criteria = list()
   designcounter = 1
 
-  for (i in seq_len(repeats)) {
+  for (i in seq_len(length(genOutput))) {
     if (!is.na(genOutput[[i]]["criterion"])) {
       designs[designcounter] = genOutput[[i]]["modelmatrix"]
       rowindicies[designcounter] = genOutput[[i]]["indices"]

R/plot_correlations.R

@@ -11,6 +11,7 @@
 #'@param custompar Default NULL. Custom parameters to pass to the `par` function for base R plotting.
 #'@param standardize Default `TRUE`. Whether to standardize (scale to -1 and 1 and center) the continuous numeric columns. Not
 #'standardizing the numeric columns can increase multi-collinearity (predictors that are correlated with other predictors in the model).
+#'@param plot Default `TRUE`. If `FALSE`, this will return the correlation matrix.
 #'@return Silently returns the correlation matrix with the proper row and column names.
 #'@import graphics grDevices
 #'@export
@@ -34,7 +35,7 @@
 #'plot_correlations(cardesign, customcolors = c("blue", "grey", "red"))
 #'plot_correlations(cardesign, customcolors = c("blue", "green", "yellow", "orange", "red"))
 plot_correlations = function(genoutput, model = NULL, customcolors = NULL, pow = 2, custompar = NULL,
-                             standardize = TRUE) {
+                             standardize = TRUE, plot = TRUE) {
   #Remove skpr-generated REML blocking indicators if present
   if (!is.null(attr(genoutput, "splitanalyzable"))) {
     if (attr(genoutput, "splitanalyzable")) {
@@ -107,6 +108,9 @@ plot_correlations = function(genoutput, model = NULL, customcolors = NULL, pow =
   mm = model.matrix(model, genoutput, contrasts.arg = contrastlist)
   #Generate pseudo inverse as it's likely the model matrix will be singular with extra terms
   cormat = abs(cov2cor(getPseudoInverse(t(mm) %*% solve(V) %*% mm))[-1, -1])
+  if(!plot) {
+    return(cormat)
+  }
 
   if (is.null(customcolors)) {
     imagecolors = viridis::viridis(101)

src/genOptimalDesign.cpp

@@ -38,12 +38,12 @@ List genOptimalDesign(Eigen::MatrixXd initialdesign, const Eigen::MatrixXd& cand
   Eigen::MatrixXd test(initialdesign.cols(), initialdesign.cols());
   test.setZero();
   if(nTrials < candidatelist.cols()) {
-    throw std::runtime_error("skpr: Too few runs to generate initial non-singular matrix: increase the number of runs or decrease the number of parameters in the matrix");
+     Rcpp::stop("skpr: Too few runs to generate initial non-singular matrix: increase the number of runs or decrease the number of parameters in the matrix");
   }
   //Check for singularity from a column perfectly correlating with the intercept.
   for(int j = 1; j < candidatelist.cols(); j++) {
     if(candidatelist.col(0).cwiseEqual(candidatelist.col(j)).all()) {
-      throw std::runtime_error("skpr: Singular model matrix from factor aliased into intercept, revise model");
+      Rcpp::stop("skpr: Singular model matrix from factor aliased into intercept, revise model");
     }
   }
   Eigen::VectorXi shuffledindices;

src/genSplitPlotOptimalDesign.cpp

@@ -44,7 +44,7 @@ List genSplitPlotOptimalDesign(Eigen::MatrixXd initialdesign, Eigen::MatrixXd ca
   //Checks if the initial matrix is singular. If so, randomly generates a new design nTrials times.
   for(int j = 1; j < candidatelist.cols(); j++) {
     if(ones.col(0).cwiseEqual(candidatelist.col(j)).all()) {
-      throw std::runtime_error("skpr: Singular model matrix from factor aliased into intercept, revise model");
+      Rcpp::stop("skpr: Singular model matrix from factor aliased into intercept, revise model");
     }
   }
   int nTrials = initialdesign.rows();
@@ -82,7 +82,7 @@ List genSplitPlotOptimalDesign(Eigen::MatrixXd initialdesign, Eigen::MatrixXd ca
   Eigen::VectorXi candidateRow = initialRows;
   Eigen::VectorXi shuffledindices;
   if(nTrials < candidatelist.cols() + blockedCols + numberinteractions) {
-    throw std::runtime_error("skpr: Too few runs to generate initial non-singular matrix: increase the number of runs or decrease the number of parameters in the matrix");
+    Rcpp::stop("skpr: Too few runs to generate initial non-singular matrix: increase the number of runs or decrease the number of parameters in the matrix");
   }
   //Checks if the initial matrix is singular. If so, randomly generates a new design maxSingularityChecks times.
   for (int check = 0; check < maxSingularityChecks; check++) {

src/getBlockedOptimalDesign.cpp

@@ -37,12 +37,12 @@ List genBlockedOptimalDesign(Eigen::MatrixXd initialdesign, const Eigen::MatrixX
   const Eigen::MatrixXd vInv = V.colPivHouseholderQr().inverse();
 
   if(nTrials < candidatelist.cols()) {
-    throw std::runtime_error("skpr: Too few runs to generate initial non-singular matrix: increase the number of runs or decrease the number of parameters in the matrix");
+    Rcpp::stop("skpr: Too few runs to generate initial non-singular matrix: increase the number of runs or decrease the number of parameters in the matrix");
   }
   //Check for singularity from a column perfectly correlating with the intercept.
   for(int j = 1; j < candidatelist.cols(); j++) {
     if(candidatelist.col(0).cwiseEqual(candidatelist.col(j)).all()) {
-      throw std::runtime_error("skpr: Singular model matrix from factor aliased into intercept, revise model");
+      Rcpp::stop("skpr: Singular model matrix from factor aliased into intercept, revise model");
     }
   }
   Eigen::VectorXi shuffledindices;