- incorporated both pseudo R square AND goodness of fit indicator in …

…the mds objects

R/ggplots.R

@@ -129,6 +129,7 @@ ggplotSamplesMDS <- function(
     #browser()
 
     RSq <- mdsObj$RSq[nDim]
+    GoF <- mdsObj$GoF[nDim]
 
     # margRSq <- rep(0., nDim)
     # 
@@ -153,28 +154,28 @@ ggplotSamplesMDS <- function(
     DF$stress <- mdsObj$spp
 
     xlabel <- paste0("Coord. ", projectionAxes[1])
-    #if (projectionAxes[1] > 2) {
-        xlabel <- paste0(
-            xlabel, 
-            #" (marg. R2 : ", 
-            #round(100*margRSq[projectionAxes[1]], 2), 
-            " (% var. : ",
-            round(100*explVar[projectionAxes[1]], 2),
-            "%)")
-    #}
+
+    xlabel <- paste0(
+        xlabel, 
+        #" (marg. R2 : ", 
+        #round(100*margRSq[projectionAxes[1]], 2), 
+        " (% var. : ",
+        round(100*explVar[projectionAxes[1]], 2),
+        "%)")
 
     ylabel <- paste0("Coord. ", projectionAxes[2])
-    #if (projectionAxes[2] > 2) {
-        ylabel <- paste0(
-            ylabel, 
-            #" (marg. R2 : ",
-            #round(100*margRSq[projectionAxes[2]], 2), 
-            " (% var. : ",
-            round(100*explVar[projectionAxes[2]], 2),
-            "%)")
-    #}
-
-    subtitle <- paste0("(Goodness of fit = ", round(RSq, 4),"; nDim = ", nDim, ")")
+
+    ylabel <- paste0(
+        ylabel, 
+        #" (marg. R2 : ",
+        #round(100*margRSq[projectionAxes[2]], 2), 
+        " (% var. : ",
+        round(100*explVar[projectionAxes[2]], 2),
+        "%)")
+
+    subtitle <- paste0("(Pseudo RSq = ", round(RSq, 4),
+                       "; Goodness of Fit = ", round(GoF, 4),
+                       "; nDim = ", nDim, ")")
 
     mainAesMapping <- ggplot2::aes(
         x = .data[["x"]],
@@ -400,8 +401,10 @@ ggplotSamplesMDSShepard <- function(
     }
 
     RSq <- mdsObj$RSq[nDim]
-    subtitle <- paste0("(Goodness of fit = ", round(RSq, 4),"; nDim = ", nDim, ")")
-    subtitle <- paste0(subtitle,")")
+    GoF <- mdsObj$GoF[nDim]
+    subtitle <- paste0("(Pseudo RSq = ", round(RSq, 4),
+                       "; Goodness of Fit = ", round(GoF, 4),
+                       "; nDim = ", nDim, ")")
 
     xlabel <- "HD distances"
     ylabel <- "Proj. distances"

R/stats.R

@@ -573,7 +573,14 @@ getChannelsSummaryStat <- function(
 #' - `spp` the stress per point obtained from the SMACOF algorithm, i.e.
 #' the contribution of each point to the stress loss function
 #' - `$RSq` R squares, for each d, from 1 to `nDim`:
-#' the R square when taking all dims from 1 to d.
+#' the (pseudo) R square when taking all dims from 1 to d.
+#' - `$GoF` Goodness of fit, for each d, from 1 to `nDim`:
+#' the goodness of fit indicator (b/w 0 and 1) when taking all dims from 1 to d.
+#' Note pseudo R square and goodness of fit indicators are essentially the 
+#' same indicator, only the definition of total sum of squares differ:
+#' - for pseudo RSq: TSS is calculated using the mean pairwise distance 
+#' as minimum
+#' - for goodness of fit: TSS is calculated using 0 as minimum
 #' 
 #' @importFrom stats as.dist dist
 #' @export
@@ -683,7 +690,10 @@ computeMetricMDS <- function(
     scaleFactor <- sqrt(sum(delta^2, na.rm = TRUE)) / sqrt(N)
     proj <- proj * scaleFactor
 
-    computeRSquares <- function(distances, projections) {
+    computeRSquares <- function(
+        distances, 
+        projections,
+        asInLinearRegression = TRUE) {
         delta <- as.dist(distances)
         nDim <- ncol(projections)
 
@@ -697,8 +707,13 @@ computeMetricMDS <- function(
             },
             FUN.VALUE = 0.)
 
-        #TSS <- sum((delta - mean(delta, na.rm = TRUE))^2)
-        TSS <- sum(delta^2)
+        if (asInLinearRegression) {
+            TSS <- sum((delta - mean(delta, na.rm = TRUE))^2)    
+        } else {
+            TSS <- sum(delta^2)    
+        }
+
+
 
         RSq <- 1-RSS/TSS
 
@@ -707,14 +722,20 @@ computeMetricMDS <- function(
 
     RSq <- computeRSquares(
         distances = pwDist,
-        projections = proj)
+        projections = proj, 
+        asInLinearRegression = TRUE)
+    GoF <- computeRSquares(
+        distances = pwDist,
+        projections = proj, 
+        asInLinearRegression = FALSE)
 
     res$pwDist <- as.dist(pwDist)
     res$proj <- proj
     res$projDist <- dist(proj)
     res$stress <- mdsRes$stress
     res$spp <- mdsRes$spp
     res$RSq <- RSq
+    res$GoF <- GoF
     res$mdsObj <- mdsRes
 
     class(res) <- "mdsRes"