new updates 16/11

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: ClusROC
 Type: Package
 Title: ROC Analysis in Three-Class Classification Problems for Clustered Data
-Date: 2022-11-15
+Date: 2022-11-16
 Version: 1.0.2
 Author: Duc-Khanh To [aut, cre] (<https://orcid.org/0000-0002-4641-0764>), with contributions from Gianfranco Adimari and Monica Chiogna
 Maintainer: Duc-Khanh To <[email protected]>
@@ -13,13 +13,14 @@ Description: Statistical methods for ROC surface analysis in three-class classif
   Methods considered in points (i), (ii) and (iv) are proposed and discussed in To et al. (2022) <doi:10.1177/09622802221089029>. Referring to point (iv), three  different selection criteria are implemented: Generalized Youden Index (GYI), Closest to Perfection (CtP) and Maximum Volume (MV). Methods considered in point (iii) are proposed and discussed in Xiong et al. (2018) <doi:10.1177/0962280217742539>. Visualization tools are also provided. We refer readers to the articles cited above for all details. 
 License: GPL-3
 Depends: R (>= 3.5.0), stats, utils, graphics, nlme, Rcpp (>= 0.12.3)
-Imports: rgl, car, numDeriv, ggplot2, ggpubr, snow, doSNOW, foreach
+Imports: rgl, ellipse, numDeriv, ggplot2, ggpubr, snow, doSNOW, foreach
 LinkingTo: Rcpp, RcppArmadillo
 Encoding: UTF-8
 LazyData: true
 LazyLoad: yes
 ByteCompile: yes
 RoxygenNote: 7.2.2
+NeedsCompilation: yes
 URL: https://github.com/toduckhanh/ClusROC
 BugReports: https://github.com/toduckhanh/ClusROC/issues
 Suggests: 

NAMESPACE

@@ -13,7 +13,6 @@ export(clus_opt_thres3)
 export(clus_roc_surface)
 export(clus_tcfs)
 export(clus_vus)
-import(car)
 import(doSNOW)
 import(foreach)
 import(ggplot2)
@@ -27,4 +26,5 @@ import(snow)
 import(stats)
 import(utils)
 importFrom(Rcpp,evalCpp)
+importFrom(ellipse,ellipse)
 useDynLib(ClusROC, .registration = TRUE)

R/clus_opt_thresh_3class.R

@@ -3,8 +3,9 @@
 ####========================================================================####
 
 #' @import grDevices
-#' @import car
+#' @import graphics
 #' @import ggplot2
+#' @importFrom ellipse ellipse
 #'
 clus_opt_thres3_core <- function(method, para, z, n_p, n_coef, boxcox, start,
                                  method_optim, maxit, lower, upper) {
@@ -351,7 +352,8 @@ clus_opt_thres3 <- function(method = c("GYI", "CtP", "MV"), out_clus_lme,
 #'@export
 plot.clus_opt_thres3 <- function(x, ci_level = 0.95, colors = NULL, xlims,
                                  ylims, size_point = 0.5, size_path = 0.5,
-                                 names_labels, nrow_legend = 1, file_name = NULL, ...) {
+                                 names_labels, nrow_legend = 1,
+                                 file_name = NULL, ...) {
   if (isFALSE(inherits(x, "clus_opt_thres3"))) {
     stop("x was not from clus_opt_thres3()!")
   }
@@ -372,55 +374,60 @@ plot.clus_opt_thres3 <- function(x, ci_level = 0.95, colors = NULL, xlims,
     theme_bw() +
     theme(legend.position = "bottom", strip.text.x = element_text(size = 9))
   if (!is.null(x$vcov_thres3)) {
-    c_alp <- sqrt(qchisq(ci_level, 2))
     dt_thres_list <- split(dt_thres, dt_thres$method)
     dt_ell_thres <- list()
     if ("GYI" %in% x$method) {
       dt_ell_thres_gyi <- data.frame()
       for (i in 1:n_x){
         uu <- ellipse(
-          center = as.numeric(dt_thres_list$`Generalized Youden Index`[i, 1:2]),
-          shape = x$vcov_thres3[[i]]$cov_gyi, radius = c_alp, draw = FALSE)
+          x = cov2cor(x$vcov_thres3[[i]]$cov_gyi)[1, 2],
+          scale = sqrt(diag(x$vcov_thres3[[i]]$cov_gyi)),
+          centre = as.numeric(dt_thres_list$`Generalized Youden Index`[i, 1:2]),
+          level = ci_level)
         dt_ell_thres_gyi <- rbind(dt_ell_thres_gyi,
                                   data.frame(uu, pts = as.factor(i)))
       }
       dt_ell_thres_gyi$method <- factor(rep(c("Generalized Youden Index"),
-                                            52 * n_x),
+                                            100 * n_x),
                                         levels = c("Generalized Youden Index"))
       dt_ell_thres$gyi <- dt_ell_thres_gyi
     }
     if ("CtP" %in% x$method) {
       dt_ell_thres_ctp <- data.frame()
       for (i in 1:n_x) {
         uu <- ellipse(
-          center = as.numeric(dt_thres_list$`Closest to Perfection`[i, 1:2]),
-          shape = x$vcov_thres3[[i]]$cov_ctp, radius = c_alp, draw = FALSE)
+          x = cov2cor(x$vcov_thres3[[i]]$cov_ctp)[1, 2],
+          scale = sqrt(diag(x$vcov_thres3[[i]]$cov_ctp)),
+          centre = as.numeric(dt_thres_list$`Closest to Perfection`[i, 1:2]),
+          level = ci_level)
         dt_ell_thres_ctp <- rbind(dt_ell_thres_ctp,
                                   data.frame(uu, pts = as.factor(i)))
       }
       dt_ell_thres_ctp$method <- factor(rep(c("Closest to Perfection"),
-                                            52 * n_x),
+                                            100 * n_x),
                                         levels = c("Closest to Perfection"))
       dt_ell_thres$ctp <- dt_ell_thres_ctp
     }
     if ("MV" %in% x$method) {
       dt_ell_thres_mv <- data.frame()
       for (i in 1:n_x) {
         uu <- ellipse(
-          center = as.numeric(dt_thres_list$`Max Volume`[i, 1:2]),
-          shape = x$vcov_thres3[[i]]$cov_mv,
-          radius = c_alp, draw = FALSE)
+          x = cov2cor(x$vcov_thres3[[i]]$cov_mv)[1, 2],
+          scale = sqrt(diag(x$vcov_thres3[[i]]$cov_mv)),
+          centre = as.numeric(dt_thres_list$`Max Volume`[i, 1:2]),
+          level = ci_level)
         dt_ell_thres_mv <- rbind(dt_ell_thres_mv,
                                  data.frame(uu, pts = as.factor(i)))
       }
-      dt_ell_thres_mv$method <- factor(rep(c("Max Volume"), 52 * n_x),
+      dt_ell_thres_mv$method <- factor(rep(c("Max Volume"), 100 * n_x),
                                        levels = c("Max Volume"))
       dt_ell_thres$mv <- dt_ell_thres_mv
     }
     dt_ell_thres <- do.call(rbind, dt_ell_thres)
     pp <- pp +
       geom_path(data = dt_ell_thres,
-                aes_string(x = "x", y = "y", group = "pts"), size = size_path)
+                aes_string(x = "x", y = "y", group = "pts"),
+                linewidth = size_path)
   }
   if (is.null(colors)) {
     colors <- topo.colors(n_x)

R/clus_roc_surface.R

@@ -1,10 +1,9 @@
 ####========================================================================####
-## This file consists of functions for estimating and ploting the ROC surface ##
+## This file consists of functions for plotting the ROC surface               ##
 ####========================================================================####
 
 #' @import rgl
 #' @import utils
-#' @import graphics
 
 shade_ellips <- function(orgi, sig, lev) {
   t1 <- sig[2, 2]

R/data_EnergyEthiopia.R

@@ -2,7 +2,7 @@
 #' @name EnergyEthiopia
 #' @docType data
 #'
-#' @description A subset of energy choice data used in Alem et al. (2016). The authors are used the full dataset to investigate the determinants of household cooking fuel choice and energy transition in urban Ethiopia. A full data is publicly available at \url{https://doi.org/10.1016/j.eneco.2016.06.025}.
+#' @description A subset of energy choice data used in Alem et al. (2016). The authors are used the full dataset to investigate the determinants of household cooking fuel choice and energy transition in urban Ethiopia. A full data is publicly available at \doi{10.1016/j.eneco.2016.06.025}.
 #'
 #' @usage data(EnergyEthiopia)
 #'

README.md

@@ -9,7 +9,7 @@ This package implements the techniques for ROC surface analysis, in cases of clu
   - [x] the optimal pairs of thresholds;
   - [x] the volume under ROC surface (VUS).
 
-Methods considered for the first three quantities are proposed and discussed in [To et al. (2022)](https://journals.sagepub.com/doi/full/10.1177/09622802221089029) . Referring to the third, three  different selection criteria are implemented: Generalized Youden Index (GYI), Closest to Perfection (CtP) and Maximum Volume (MV). Methods considered for the last are proposed and discussed in [Xiong et al. (2018)](https://journals.sagepub.com/doi/full/10.1177/0962280217742539). Visualization tools are also provided. We refer readers to the articles cited above for all details.
+Methods considered for the first three quantities are proposed and discussed in [To et al. (2022)](https://doi.org/10.1177/09622802221089029). Referring to the third, three different selection criteria are implemented: Generalized Youden Index (GYI), Closest to Perfection (CtP) and Maximum Volume (MV). Methods considered for the last are proposed and discussed in [Xiong et al. (2018)](https://doi.org/10.1177/0962280217742539). Visualization tools are also provided. We refer readers to the articles cited above for all details.
 
 ## INSTALLATION