Remove data from dataset

[alex7tula]

In dataset.R
Instead

# Remove data from dataset if necessary
# TODO better name?
postProcessDataSet <- function(dataSet = get("dataSet", envir = parent.frame()),
  darch = get("darch", envir = parent.frame()))
{
  if (!getParameter(".retainData", F))
  {
    dataSet@data <- NULL
    dataSet@targets <- NULL
  }
  
  dataSet
}

I offer:

# Remove data from dataset if necessary
# TODO better name?
postProcessDataSet <- function(dataSet = get("dataSet", envir = parent.frame()),
  darch = get("darch", envir = parent.frame()))
{
  if (!getParameter(".retainData", F))
  {
    dataSet@data <- NULL
    dataSet@targets <- NULL
    dataSet@parameters<-list();
  }
  
  dataSet
}

dataSet@parameters<-list(); - will clear educate data|matrix. With this i lose file from 2 mB to 50 kB.

But I'm not sure, maybe these data are needed somewhere?

[saviola777]

Thanks for your feedback. Yes, this huge parameters list has been a thorn in my side for a while (I do not generate it myself, it is generated by the caret library), but removing it removes information necessary for pre-processing. So after you remove it, the network may not be able to predict new data if they needed to be pre-processed.

This is currently implemented for autosaving, if you pass the parameter autosave.trim – this will reset both the dataset and the network itself.

I will keep this open for now, maybe I can find a way of reducing the size of this list without breaking pre-processing, or maybe I can add a parameter to give the user a choice of removing these parameters when they are not needed.