new code for plotting svm accuracy on test data

analysis/Rcode/analysis_for_paper1/SVM_logreg_plotresults.R

@@ -0,0 +1,46 @@
+# plot data from svm against logreg experiment
+
+load("C:/Users/cogneuro/Desktop/HCS_analysis/analysis/svm_logreg_174")
+Plothist_bin <- function (Acc_cumm,i){
+  dataplot = data.frame(Acc_cumm [-1,i])
+  realacc=Acc_cumm [1,i]
+  title = names (Acc_cumm)[i]
+  names (dataplot) = "Accuracy_perm_grouping"
+
+  k1 <- sum(as.numeric(dataplot$Accuracy_perm_grouping) >=
+              realacc)   # One-tailed test
+  stat =print(zapsmall(binconf(k1, nrow(dataplot)-1, method='exact'))) # 95% CI by default
+  R= "not enough \npermutation done"
+  if (stat[1,2] > 0.05) R= "no difference spotted"
+  if (stat[1,3] < 0.05) R= paste0("different \nwith p < ",stat[1,3])
+
+
+  p=  ggplot(data=dataplot, aes(dataplot$Accuracy_perm_grouping)) + 
+    geom_histogram(aes(y =..density..), 
+                   binwidth=0.09)+
+    #xlab (paste0 ("accuracy of grouping with ",nrow(dataplot)," permutations performed"))  +
+    xlim (c(-1,1))+
+    #geom_density(col=3) +
+    geom_vline(xintercept=realacc, col=2)+
+    labs(title= title,
+         x="",#paste0 ("accuracy of grouping with ",nrow(dataplot)," permutations performed"),
+         y="Proportion")+
+    annotate("text", label = R, x = -1, y = 1, size = 4, colour = 1,hjust = 0)
+  return(p)
+}
+
+#---- prepare plots
+Nplot <- ncol(Acc_cumm) 
+pl_hist <- list()
+
+for (i in seq(1:Nplot)){
+  pl_hist [[i]] = Plothist_bin(Acc_cumm,i)
+}
+
+#pdf(file = paste0("svm_logregresults.pdf"), width = 15, height = 10)
+
+marrangeGrob(pl_hist ,
+             ncol=3,nrow =3, top= paste0 ("accuracy of grouping with ",nrow(dataplot)," permutations performed"))
+
+
+#dev.off()