mapping caret - xgb multiclass classification

R-package/DESCRIPTION

@@ -5,7 +5,8 @@ Version: 0.1-1
 Date: 2015-08-03
 Author: Gino Tesei <gtesei@yahoo.com>
 Maintainer: Gino Tesei <gtesei@yahoo.com>
-Description: fast-furiuos gathers code, models and meta-models I needed in my Machine Learning Lab but I didn't found on the shelf.
+Description: fast-furiuos gathers code, models and meta-models I needed in my
+    Machine Learning Lab but I didn't found on the shelf.
 License: MIT + file LICENSE
 URL: https://github.com/gtesei/fast-furious
 BugReports: https://github.com/gtesei/fast-furious/issues
@@ -14,25 +15,26 @@ Suggests:
     knitr,
     lattice (>= 0.20),
     ggplot2 (>= 1.0.0),
-    testthat, 
+    testthat,
     Cubist,
-    arm, 
+    arm,
     MASS,
     kknn,
     kernlab,
-    ipred, 
-    randomForest, 
+    ipred,
+    randomForest,
     pROC
 Depends:
-    R (>= 2.10),  
+    R (>= 2.10),
     caret
 Imports:
-    parallel, 
+    parallel,
     subselect,
-    plyr, 
-    xgboost, 
-    magrittr, 
+    plyr,
+    xgboost,
+    magrittr,
     stringr,
-    e1071, 
-    glmnet, 
-    verification
+    e1071,
+    glmnet,
+    verification
+RoxygenNote: 5.0.1

R-package/NAMESPACE

@@ -1,4 +1,4 @@
-# Generated by roxygen2 (4.1.1): do not edit by hand
+# Generated by roxygen2: do not edit by hand
 
 export(RMSE.xgb)
 export(RMSLE.xgb)

R-package/R/fastClassification.R

@@ -96,6 +96,8 @@ ff.trainAndPredict.class = function(Ytrain ,
       ret = NULL
       if (metric == "auc") {
         ret = "ROC"
+      } else if (metric == "merror" | metric == "error" ) {
+        ret = "Accuracy"
       } else {
         stop(paste0("unrecognized metric:",metric))
       }