PRIM.real.data.min {rfcdmin}R Documentation

PRIM output from VRCmin data analysis

Description

The HIV-protein stimulated st.1829 FCS R-object and the unstimulated unst.1829 FCS R-object from the VRCmin data were used to exemplify the PRIM algorithm on a real dataset. The three steps for the finding of a single rule were saved as output. The out.peel is the 'PRIM.step' class object of the Peeling step. The out.expand is the 'PRIM.step' class object of the Expansion Step. The out.cv is the 'PRIM.crossval.step' class object of the cross-validation step. From these output objects we obtain the necessary indices and vectors for making plots that are exemplified in PRIM.pdf.

Usage

data(PRIM.real.data.min)

Format

box.idx.list
Peeling Step: a list of five boxes that were in the peeling sequence of 'out.peel' which used the whole data 'X'

box.support.PS
Peeling Step: The vector of proportions of X that lie in the peeled box for each iteration of the expansion step from 'out.peel'
means.PS
Peeling Step: The vector of means(Y) of each peeled box for each iteration from 'out.peel'
type.PS
Peeling Step: The vector of peeling types, denoting the direction and variable for the addition of new points into the box for each iteration from 'out.peel'
box.support.ES
Expansion Step: The vector of proportions of X that lie in the expanded box for each iteration of the expansion step from 'out.expand'
means.ES
Expansion Step: The vector of means(Y) of each expanded box for each iteration from 'out.expand'
type.ES
Expansion Step: The vector of expansion types, denoting the direction and variable for the addition of new points into the box for each iteration from 'out.expand'
best.box.X.ranges
Expansion Step: The X-variable ranges of the final rule estimated after the Expansion Step from 'out.expand'
Rule1.idx
Expansion Step: The index of the final rule obtained after the expansion step from out.expand
TD1
Cross-Validation Step: vector of positional row indices corresponding to the first testdata set which is subset from the original 'X' data (which is constructed in the example below)
TD2
Cross-Validation Step: vector of positional row indices corresponding to the second testdata set which is subset from the original 'X' data (which is constructed in the example below)
box.idx.listTD1
Cross-Validation Step: a list of five boxes in the peeling sequence of the first Testdata set
box.idx.listTD2
Cross-Validation Step: a list of five boxes in the peeling sequence of the second Testdata set

Details

The class information, extraction, plotting, implementation tools for the 'PRIM.step' and the 'PRIM.crossval.step' class objects are detailed in the rfcprim package.

Source

See VRCmin

References

See VRCmin and rfcprim.

Examples

data(PRIM.real.data)

## the following code was used to generate the output
if (FALSE){

  data(VRCmin)
  ## the HIV-protein stimulation status
  Y <- c(rep(1, dim.FCS(st.1829)[1]), 
       rep(0, dim.FCS(unst.1829)[1]))

  ## the dataset
  X <- rbind(as(st.1829, "matrix"), as(unst.1829, "matrix"))
  if (require(rfcprim)){
  out.peel <- peel.step(X,
                      Y,
                      min.box.size = 500,
                      alpha=0.10,
                      verbose=TRUE)

  out.expand <- expand.step(X,
                          Y,
                          out.peel@best.box.idx,
                          beta=0.01,
                          verbose=TRUE)

  out.cv <- crossval.step(X,
                        Y,
                        num.testdata=2,
                        prob.testdata=c(0.50, 0.50),
                        alpha=0.10,
                        target.mu.Y=1,
                        min.box.size=500,
                        choose.best.box.decision=c("max.mean.box", "final.box"),
                       
                        beta=0.01,
                        verbose=TRUE)

  box.idx.list <- list(out.peel@best.box.idx,
     out.expand@best.box.idx)

  box.support.PS <- out.peel@box.support.vec

  means.PS <- out.peel@means.vec

  type.PS <- out.peel@type.vec

  box.support.ES <- out.expand@box.support.vec

  means.ES <- out.expand@means.vec

  type.ES <- out.expand@type.vec

  best.box.X.ranges <- out.expand@best.box.X.ranges

  Rule1.idx <- out.expand@best.box.idx

  TD1 <- out.cv[1, "testdata.idx.list"]
  TD2 <- out.cv[2, "testdata.idx.list"]

  box.idx.listTD1 <- list(out.cv[1,"step.set.list"]@peel.step@best.box.idx,
      out.cv[1, "step.set.list"]@expand.step@best.box.idx)

  box.idx.listTD2 <- list(out.cv[2,"step.set.list"]@peel.step@best.box.idx,
      out.cv[2, "step.set.list"]@expand.step@best.box.idx)
  }
 }


[Package rfcdmin version 1.6.5 Index]