polished the idframe slicing function

1 files changed, 19 insertions, 43 deletions

diff --git a/R/partition.R b/R/partition.R
index 7c0af2b..7e1cc01 100644
--- a/R/partition.R
+++ b/R/partition.R
@@ -1,60 +1,36 @@
 #' Subset an idframe data
 #' 
-#' Subsetting method for datas of class \code{idframe}
+#' Subsetting method for objects of class \code{idframe}
 #' 
 #' @param data an object of class \code{idframe}
-#' @param indices the indices that need to be subsetted
+#' @param start the start
 #' @export
-dataSlice <- function(data,indices){
+dataSlice <- function(data,start=NULL,end=NULL,freq=NULL){
   # check if the class is correct
   if(class(data)!='idframe')
     stop("Not an idframe data")
   
-  if(!all(indices %in% seq(to=dim(data$output)[1],by=1)))
-    stop("Invalid indices")
+  nin <- dim(data$input)[2]; nout <- dim(data$output)[2]
+  dataMatrix <- cbind(data$input,data$output)
+  if(type=="freq"){
+    dataMatrix <- cbind(dataMatrix,data$frequencies)
+  } else {
+    timeSeq <- seq(from=data$t.start,to=data$t.end,by=data$Ts)
+    dataMatrix <- cbind(dataMatrix,timeSeq)
+  }
   
-  trim <- data
-  trim$output <- trim$output[indices,,drop=F]
-  trim$input <- trim$input[indices,,drop=F]
+  l <- as.list(dataMatrix)
+  trimData <- as.data.frame(sapply(l,window,start=start,end=end,deltat=freq))
+  
+  trim <- idframe(output=trimData[,(nin+1):(nin+nout+1)],input=trimData[,1:nin],
+                  type=data$type,Ts=data$tTs,tUnit=data$tUnit)
   
   if(trim$type=="freq"){
-    trim$frequncies <- trim$frequencies[indices] 
+    trim$frequncies <- trimData[,ncol(trimData)]
   } else {
-    trim$t.start <- trim$t.start + trim$Ts*(indices[1]-1)
-    trim$t.end <- trim$t.start + trim$Ts*(length(indices)-1)
+    trim$t.start <- trimData[1,ncol(trimData)]
+    trim$t.end <- trimData[nrow(trimData),ncol(trimData)]
   }
   
   return(trim)
-}
-
-#' Split data into training and validation sets
-#' 
-#' The function splits the data into training and validation sets and returns them bundled
-#' as a list. The size of the sets are determined by the parameter \code{p}.
-#' 
-#' @param data an object of class \code{idframe}
-#' @param p the percentage of the data that goes to training (Default : \code{0.6})
-#' @return list containing estimation and validation idframe datas
-#' 
-#' @examples
-#' data(cstr)
-#' splitList <- dataPartition(cstr,p=0.6)
-#' train <- splitList$estimation # training set 
-#' test <- splitList$validation # testing set
-#' 
-#' @export
-dataPartition <- function(data,p=0.6){
-  # check if the class is correct
-  if(class(data)!='idframe')
-    stop("Not an idframe data")
-  
-  index <- seq_along(data$output[,1])
-  
-  trainIndex <- index[1:round(p*length(index))]
-  testIndex <- index[!(index %in% trainIndex)]
-  
-  train <- dataSlice(data,trainIndex)
-  test <- dataSlice(data,testIndex)
-  
-  return(list(estimation=train,validation=test))
 }
\ No newline at end of file