improve test on objects

R/kafino.R

@@ -474,10 +474,8 @@ kafino_fit<-function(datain,Tvar,Yvar,
       resultat=FK_para_connu_tronc(param)
       
     } else {
-      # paramètre de troncature, dépend du nombre d'observation à étudier
-      if (N <= 5) {
-        print("Warning: not enough data between expert knowledge. The algorithm is not performed.")
-      } else {
+      # N petit, si trop petit on ne fait rien
+      if (N > 5) {
         # Not enough data - no optim
         print("-------:")
         print("Optimisation of initial parameters - result:")
@@ -485,35 +483,42 @@ kafino_fit<-function(datain,Tvar,Yvar,
         print(X)
         print("-------:")
         resultat=FK_para_connu_tronc(X)
+      } else {
+        print("Warning: not enough data between expert knowledge. The algorithm is not performed.")
+        resultat<-NULL
       }
     }
   } else {
-    if (N <= 5) {
-      print("Warning: not enough data between expert knowledge. The algorithm is not performed.")
-      resultat<-NULL
-    } else {
+    # Pas d'optimisation et test si N petit - si trop petit on ne fait rien
+    if (N > 5) {
       # No optimisation on initial parameters
       print("-------:")
       print("No optimisation of initial parameters:")
       print("Used parameters: ")
       print(X)
       resultat=FK_para_connu_tronc(X)
+    } else {
+      print("Warning: not enough data between expert knowledge. The algorithm is not performed.")
+      resultat<-NULL
     }
   }
   
   #--------------------------------
   # Formatting output results
+  # If resultat NULL then create output with datain and 2 NULL objects
+  # else create output with detectoutlier, PredictionOK and kafino.results
+  # useful for the kafino_plot() function
   #--------------------------------
   if (is.null(resultat)){
     dt.out<-datain
     dt.pred<-NULL
     resultat<-NULL    
+    
     mylist<-list(dt.out,dt.pred,resultat)
     names(mylist)<-c("detectOutlier","PredictionOK","kafino.results")
     class(mylist) = c("kafino")
     return(invisible(mylist))
-    
-  } else{
+  } else {
     prediction=na.omit(resultat$prediction)
     label_pred=round(na.omit(resultat$label),2)
     lwr=na.omit(resultat$lwr)