diff --git a/DESCRIPTION b/DESCRIPTION
index 92c2ebe..948b9c9 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -15,7 +15,7 @@ Suggests:
     randomForest,
     caret
 Depends:
-    R (>= 3.2.0),
+    R (>= 3.0.0),
     methods
 Collate:
     'VirtualTwins.R'
diff --git a/R/predict.R b/R/predict.R
index b42d233..76d4bfd 100644
--- a/R/predict.R
+++ b/R/predict.R
@@ -15,9 +15,12 @@
 #'   
 #' @return vector \eqn{E(Y=1)}
 #'   
+#'
 #' @include setClass.R
 #' @importClassesFrom party RandomForest
 #' 
+#' @name VT.predict
+#' 
 setGeneric("VT.predict",
            function(rfor, newdata, type){standardGeneric("VT.predict")}
 )
diff --git a/R/tree.R b/R/tree.R
index 58ae35d..c655c43 100644
--- a/R/tree.R
+++ b/R/tree.R
@@ -1,17 +1,47 @@
-# TREES COMPUTATIONS ------------------------------------------------------
+#' Tree to find subgroup
+#' 
 #' An abstract reference class to compute tree
 #' 
-#' @include difft.R setClass.R
+#' \code{VT.tree.class} and \code{VT.tree.reg} are children of \code{VT.tree}.
+#' \code{VT.tree.class} and \code{VT.tree.reg} try to find a strong association 
+#' between \code{difft} (in \code{VT.difft} object) and RCT variables.
 #' 
-#' @field vt.difft VT.difft object
-#' @field outcome vector
-#' @field threshold numeric Threshold for difft (c)
-#' @field screening logical TRUE if using varimp (default is VT.object screening field) 
-#' @field sens character Sens can be ">" (default) or "<". Meaning : difft > threshold or difft < threshold
+#' In \code{VT.tree.reg}, a regression tree is computed on \code{difft} values. 
+#' Then, thanks to the \code{threshold} it flags leafs of the \code{tree} which 
+#' are above the \code{threshold} (when \code{sens} is ">"). Or it flags leafs 
+#' which are below the \code{threshold} (when \code{sens} = "<").
+#' 
+#' In \code{VT.tree.class}, it first flags \code{difft} above or below 
+#' (depending on the \code{sens}) the given \code{threshold}. Then a 
+#' classification tree is computed to find which variables explain flagged 
+#' \code{difft}.
+#' 
+#' To sum up, \code{VT.tree} try to understand which variables are associated 
+#' with a big change of \code{difft}.
+#' 
+#' Results are shown with \code{getRules()} function. \code{only.leaf} parameter
+#' allows to obtain only the leaf of the \code{tree}. \code{only.fav} parameter 
+#' select only favorable nodes. \code{tables} shows incidence table of the rule.
+#' \code{verbose} allow \code{getRules()} to be quiet. And \code{compete} show
+#' also rules with \code{maxcompete} competitors from the \code{tree}.
+#' 
+#' @include difft.R setClass.R
+#'   
+#' @field vt.difft \code{VT.difft} object
+#' @field outcome outcome vector from \code{rpart} function
+#' @field threshold numeric Threshold for difft calculation (c)
+#' @field screening Logical. TRUE if using varimp. Default is VT.object 
+#'   screening field
+#' @field sens character Sens can be ">" (default) or "<". Meaning : 
+#'   \code{difft} > \code{threshold} or \code{difft} < \code{threshold}
 #' @field name character Names of the tree
 #' @field tree rpart Rpart object to construct the tree
 #' @field Ahat vector Indicator of beglonging to Ahat
-#'  
+#'   
+#' @seealso \code{\link{VT.tree.reg}}, \code{\link{VT.tree.class}}
+#'   
+#' @name VT.tree
+#'   
 #' @import methods
 VT.tree <- setRefClass(
   Class = "VT.tree",
@@ -29,7 +59,7 @@ VT.tree <- setRefClass(
   ),
   
   methods = list(
-    initialize = function(vt.difft = VT.difft(), threshold = 0.05, sens = ">", screening = NULL){        
+    initialize = function(vt.difft = VT.difft(), threshold = 0.05, sens = ">", screening = NULL){
       .self$vt.difft <- vt.difft
       
       .self$threshold <- threshold
@@ -41,6 +71,7 @@ VT.tree <- setRefClass(
     },
     
     getData = function(){
+      "Return data used for tree computation"
       d <- .self$vt.difft$vt.object$data[, 3:ncol(.self$vt.difft$vt.object$data)]
       
       if(.self$screening == T){
@@ -55,6 +86,7 @@ VT.tree <- setRefClass(
     },
     
     computeNameOfTree = function(type){
+      "return label of response variable of the tree"
       return(type)
       if(.self$threshold < 0 ){
         threshold.chr <- paste0("m", -.self$threshold)
@@ -66,11 +98,13 @@ VT.tree <- setRefClass(
       return(paste(type, tmp[1], tmp[2], sep = ""))
     },
     
-    run = function(){
+    run = function(...){
+      "Compute tree with rpart parameters"
       if(length(.self$vt.difft$difft) == 0) stop("VT.difft::difft is an empty vector")
     },
     
     getInfos = function(){
+      "Return infos about tree"
       cat("\n")
       cat(sprintf("Threshold = %0.4f", .self$threshold))
       cat("\n")
@@ -86,6 +120,7 @@ VT.tree <- setRefClass(
     },
     
     getRules = function(only.leaf = F, only.fav = F, tables = T, verbose = T, compete = F){
+      "Retrun subgroups discovered by the tree. See details."
       
       # On crée le tableau des competitors
       if(isTRUE(compete))
@@ -226,6 +261,7 @@ VT.tree <- setRefClass(
     },
     
     createCompetitors = function(){
+      "Create competitors table"
       
       fr  <- .self$tree$frame
       fr  <- fr[fr$var != "<leaf>",]
@@ -253,10 +289,12 @@ VT.tree <- setRefClass(
     },
     
     getIncidences = function(rule, rr.snd = T){
+      "Return incidence of the rule"
       return(VT.incidences(.self$vt.difft, rule, rr.snd))
     },
     
     getAhatIncidence = function(){
+      "Return Ahat incidence"
       if(sum(.self$Ahat)!=0){
         
         table.inc <- VT.incidences(vt.object = .self$vt.difft$vt.object, select = .self$Ahat)
@@ -279,6 +317,7 @@ VT.tree <- setRefClass(
     },
     
     getAhatQuality = function(){
+      "Return Ahat quality"
       
       resub <- vt.getQAOriginal(.self$Ahat, response = .self$vt.difft$vt.object$getY(), trt = .self$vt.difft$vt.object$data[, 2])
       
diff --git a/R/tree.class.R b/R/tree.class.R
index 0700a7c..37fb844 100644
--- a/R/tree.class.R
+++ b/R/tree.class.R
@@ -1,9 +1,13 @@
 # VT.TREE.CLASS -----------------------------------------------------------
 
-#' A reference class to compute subgroups by classifiation tree
+#' Classification tree to find subgroups
+#'
+#' See \code{\link{VT.tree}}
 #' 
 #' @include tree.R
 #' 
+#' @name VT.tree.class
+#' 
 #' @import methods
 VT.tree.class <- setRefClass(
   Class = "VT.tree.class",
@@ -24,6 +28,7 @@ VT.tree.class <- setRefClass(
     },
     
     run = function(...){
+      "VT.tree.class:run(...) Compute classification tree with rpart parameters"
       callSuper()
       
       data  <- .self$getData()  
diff --git a/R/tree.reg.R b/R/tree.reg.R
index 1893d72..d7dde60 100644
--- a/R/tree.reg.R
+++ b/R/tree.reg.R
@@ -1,8 +1,12 @@
 # VT.TREE.REG -------------------------------------------------------------
 
+#' Regression tree to find subgroups
+#'
+#' See \code{\link{VT.tree}}
+#'
 #' @include tree.R
-#' A reference class to compute subgroups by regression tree with rpart package
 #' 
+#' @name VT.tree.reg
 
 VT.tree.reg <- setRefClass(
   Class = "VT.tree.reg",
@@ -20,7 +24,6 @@ VT.tree.reg <- setRefClass(
     
     run = function(...){
       callSuper()
-      
       data  <- .self$getData()
       
       .self$tree <- rpart::rpart(as.formula(paste(.self$name, ".", sep = "~")), data = data, ...)
diff --git a/man/VT.difft.Rd b/man/VT.difft.Rd
index 1320639..4b36f71 100644
--- a/man/VT.difft.Rd
+++ b/man/VT.difft.Rd
@@ -20,7 +20,7 @@ favorable outcome is 1. Then, \deqn{difft_i = twin1_i - twin2_i IF T_i =
 
 \item{\code{twin1}}{vector of \eqn{E(Y|T = real treatment)}}
 
-\item{\code{twin2}}{vector of \eqn{E(Y|T = anoher treatment)}}
+\item{\code{twin2}}{vector of \eqn{E(Y|T = another treatment)}}
 
 \item{\code{difft}}{vector of difference between twin1 and twin2}
 }}
diff --git a/man/VT.forest.Rd b/man/VT.forest.Rd
index c6e832f..65a2eff 100644
--- a/man/VT.forest.Rd
+++ b/man/VT.forest.Rd
@@ -22,7 +22,7 @@ An abstract reference class to compute twin via random forests
 
 \item{\code{getFullData()}}{Return twin1, twin2 and difft in column}
 
-\item{\code{run()}}{Compute twin1 and twin2 computation. Switch treatment if necessary.}
+\item{\code{run()}}{Compute twin1 and twin2 estimation. Switch treatment if necessary.}
 }}
 \seealso{
 \code{\link{VT.difft}}, \code{\link{VT.forest.one}}, \code{\link{VT.forest.double}}
diff --git a/man/VT.forest.fold.Rd b/man/VT.forest.fold.Rd
index 07cdea6..753492b 100644
--- a/man/VT.forest.fold.Rd
+++ b/man/VT.forest.fold.Rd
@@ -31,7 +31,7 @@ interactions}
 \section{Methods}{
 
 \describe{
-\item{\code{run()}}{Compute twin1 and twin2 computation. Switch treatment if necessary.}
+\item{\code{run()}}{Compute twin1 and twin2 estimation. Switch treatment if necessary.}
 }}
 \seealso{
 \code{\link{VT.difft}}, \code{\link{VT.forest}},
diff --git a/man/VT.tree-class.Rd b/man/VT.tree-class.Rd
deleted file mode 100644
index ec4b9de..0000000
--- a/man/VT.tree-class.Rd
+++ /dev/null
@@ -1,30 +0,0 @@
-% Generated by roxygen2 (4.1.1): do not edit by hand
-% Please edit documentation in R/tree.R
-\docType{class}
-\name{VT.tree-class}
-\alias{VT.tree}
-\alias{VT.tree-class}
-\title{An abstract reference class to compute tree}
-\description{
-An abstract reference class to compute tree
-}
-\section{Fields}{
-
-\describe{
-\item{\code{vt.difft}}{VT.difft object}
-
-\item{\code{outcome}}{vector}
-
-\item{\code{threshold}}{numeric Threshold for difft (c)}
-
-\item{\code{screening}}{logical TRUE if using varimp (default is VT.object screening field)}
-
-\item{\code{sens}}{character Sens can be ">" (default) or "<". Meaning : difft > threshold or difft < threshold}
-
-\item{\code{name}}{character Names of the tree}
-
-\item{\code{tree}}{rpart Rpart object to construct the tree}
-
-\item{\code{Ahat}}{vector Indicator of beglonging to Ahat}
-}}
-
diff --git a/man/VT.tree.Rd b/man/VT.tree.Rd
new file mode 100644
index 0000000..7fe0bb4
--- /dev/null
+++ b/man/VT.tree.Rd
@@ -0,0 +1,80 @@
+% Generated by roxygen2 (4.1.1): do not edit by hand
+% Please edit documentation in R/tree.R
+\docType{class}
+\name{VT.tree}
+\alias{VT.tree}
+\title{Tree to find subgroup}
+\description{
+An abstract reference class to compute tree
+}
+\details{
+\code{VT.tree.class} and \code{VT.tree.reg} are children of \code{VT.tree}.
+\code{VT.tree.class} and \code{VT.tree.reg} try to find a strong association
+between \code{difft} (in \code{VT.difft} object) and RCT variables.
+
+In \code{VT.tree.reg}, a regression tree is computed on \code{difft} values.
+Then, thanks to the \code{threshold} it flags leafs of the \code{tree} which
+are above the \code{threshold} (when \code{sens} is ">"). Or it flags leafs
+which are below the \code{threshold} (when \code{sens} = "<").
+
+In \code{VT.tree.class}, it first flags \code{difft} above or below
+(depending on the \code{sens}) the given \code{threshold}. Then a
+classification tree is computed to find which variables explain flagged
+\code{difft}.
+
+To sum up, \code{VT.tree} try to understand which variables are associated
+with a big change of \code{difft}.
+
+Results are shown with \code{getRules()} function. \code{only.leaf} parameter
+allows to obtain only the leaf of the \code{tree}. \code{only.fav} parameter
+select only favorable nodes. \code{tables} shows incidence table of the rule.
+\code{verbose} allow \code{getRules()} to be quiet. And \code{compete} show
+also rules with \code{maxcompete} competitors from the \code{tree}.
+}
+\section{Fields}{
+
+\describe{
+\item{\code{vt.difft}}{\code{VT.difft} object}
+
+\item{\code{outcome}}{outcome vector from \code{rpart} function}
+
+\item{\code{threshold}}{numeric Threshold for difft calculation (c)}
+
+\item{\code{screening}}{Logical. TRUE if using varimp. Default is VT.object
+screening field}
+
+\item{\code{sens}}{character Sens can be ">" (default) or "<". Meaning :
+\code{difft} > \code{threshold} or \code{difft} < \code{threshold}}
+
+\item{\code{name}}{character Names of the tree}
+
+\item{\code{tree}}{rpart Rpart object to construct the tree}
+
+\item{\code{Ahat}}{vector Indicator of beglonging to Ahat}
+}}
+\section{Methods}{
+
+\describe{
+\item{\code{computeNameOfTree(type)}}{return label of response variable of the tree}
+
+\item{\code{createCompetitors()}}{Create competitors table}
+
+\item{\code{getAhatIncidence()}}{Return Ahat incidence}
+
+\item{\code{getAhatQuality()}}{Return Ahat quality}
+
+\item{\code{getData()}}{Return data used for tree computation}
+
+\item{\code{getIncidences(rule, rr.snd = T)}}{Return incidence of the rule}
+
+\item{\code{getInfos()}}{Return infos about tree}
+
+\item{\code{getRules(only.leaf = F, only.fav = F, tables = T, verbose = T,
+  compete = F)}}{Retrun subgroups discovered by the tree. See details.}
+
+\item{\code{run(...)}}{Compute tree with rpart parameters}
+}}
+\seealso{
+\code{\link{VT.tree.reg}}, \code{\link{VT.tree.class}}
+}
+
diff --git a/man/VT.tree.class-class.Rd b/man/VT.tree.class-class.Rd
deleted file mode 100644
index fcb45ba..0000000
--- a/man/VT.tree.class-class.Rd
+++ /dev/null
@@ -1,11 +0,0 @@
-% Generated by roxygen2 (4.1.1): do not edit by hand
-% Please edit documentation in R/tree.class.R
-\docType{class}
-\name{VT.tree.class-class}
-\alias{VT.tree.class}
-\alias{VT.tree.class-class}
-\title{A reference class to compute subgroups by classifiation tree}
-\description{
-A reference class to compute subgroups by classifiation tree
-}
-
diff --git a/man/VT.tree.class.Rd b/man/VT.tree.class.Rd
new file mode 100644
index 0000000..3f3939d
--- /dev/null
+++ b/man/VT.tree.class.Rd
@@ -0,0 +1,15 @@
+% Generated by roxygen2 (4.1.1): do not edit by hand
+% Please edit documentation in R/tree.class.R
+\docType{class}
+\name{VT.tree.class}
+\alias{VT.tree.class}
+\title{Classification tree to find subgroups}
+\description{
+See \code{\link{VT.tree}}
+}
+\section{Methods}{
+
+\describe{
+\item{\code{run(...)}}{Compute tree with rpart parameters}
+}}
+
diff --git a/man/VT.tree.reg.Rd b/man/VT.tree.reg.Rd
new file mode 100644
index 0000000..8485e61
--- /dev/null
+++ b/man/VT.tree.reg.Rd
@@ -0,0 +1,15 @@
+% Generated by roxygen2 (4.1.1): do not edit by hand
+% Please edit documentation in R/tree.reg.R
+\docType{class}
+\name{VT.tree.reg}
+\alias{VT.tree.reg}
+\title{Regression tree to find subgroups}
+\description{
+See \code{\link{VT.tree}}
+}
+\section{Methods}{
+
+\describe{
+\item{\code{run(...)}}{Compute tree with rpart parameters}
+}}
+