VBmix (version 0.3.2)

displayNnet:

Description

displays the colored decision regions of a neural network model. Data symbols are also optionally displayed. Data and model should be 2D.

Usage

displayNnet(nnet.model, datamatrix, datalabels, subset = NULL, 
displayPoints = TRUE, steps = 100, alpha = 0.4, lwd = 1)

Arguments

nnet.model

a neural network model, as returned by nnet (nnet library)

datamatrix

a matrix of row-elements.

datalabels

matrix of binary indicator variables for labels (as used by nnet).

subset

vector of indexes of a data subset to be displayed. If NULL, all points are displayed.

displayPoints

if FALSE, only decision regions are displayed.

steps

influences the resolution of the decision regions. Low values will provoke aliasing, high values are slower to be displayed.

alpha

alpha blending parameter between decision regions and data symbols.

lwd

magnification factor for the stroke width used to plot symbols.

Value

a new plotting window displaying decision regions associated to the parametrized neural network.

See Also

nnet

Examples

temp <- nnet::class.ind(irislabels)
temp2 <- setDomain(irisdata[,1:2], 10)
temp3 <- nnet::nnet(temp2, temp, size=10)
displayNnet(temp3, temp2, temp)