PPPDiffRef {SBCK}

PPPDiffRef

Description

Apply the diff w.r.t. a ref transformation.

Details

Transform a dataset such that all 'lower' dimensions are replaced by the 'ref' dimension minus the 'lower'; and all 'upper' dimensions are replaced by 'upper' minus 'ref'.

Super class

SBCK::PrePostProcessing -> PPPDiffRef

Public fields

ref

[integer] The reference column

lower

[vector integer] Dimensions lower than ref

upper

[vector integer] Dimensions upper than ref

Methods

Public methods

• PPPDiffRef$new()

• PPPDiffRef$transform()

• PPPDiffRef$itransform()

• PPPDiffRef$clone()

Inherited methods

• SBCK::PrePostProcessing$fit()
• SBCK::PrePostProcessing$predict()

Method new()

Create a new PPPDiffRef object.

Usage

PPPDiffRef$new(ref, lower = NULL, upper = NULL, ...)

Arguments

ref

The reference column

lower

Dimensions lower than ref

upper

Dimensions upper than ref

...

Others arguments are passed to PrePostProcessing

Returns

A new 'PPPDiffRef' object.

Method transform()

Apply the DiffRef transform.

Usage

PPPDiffRef$transform(X)

Arguments

X

Data to transform

Returns

Xt a transformed matrix

Method itransform()

Apply the DiffRef inverse transform.

Usage

PPPDiffRef$itransform(Xt)

Arguments

Xt

Data to transform

Returns

X a transformed matrix

Method clone()

The objects of this class are cloneable with this method.

Usage

PPPDiffRef$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

## Parameters
size  = 2000
nfeat = 5
sign  = base::sample( base::c(-1,1) , nfeat - 1 , replace = TRUE )

## Build data
X     = matrix( stats::rnorm( n = size ) , ncol = 1 )
for( s in sign )
{
	X = base::cbind( X , X[,1] + s * base::abs(matrix( stats::rnorm(n = size) , ncol = 1 )) )
}

## PPP
lower = which( sign ==  1 ) + 1
upper = which( sign == -1 ) + 1
ppp   = SBCK::PPPDiffRef$new( ref = 1 , lower = lower , upper = upper )
Xt    = ppp$transform(X)
Xti   = ppp$itransform(Xt)

print( base::max( base::abs( X - Xti ) ) )