Nnet: Neural Networks Using Multiple Starting Weights

In RcmdrPlugin.BCA: Rcmdr Plug-In for Business and Customer Analytics

Description

Estimates a feed forward neural network using multiple intial starting weight vectors using the nnet function, and selects as the final model the one that minimizes the criterion function. This function is designed to be used with the Rcmdrma package. The function nnSub implements subsetting in a way more analogous to other R fitting functions.

Usage

Nnet(formula, data, decay, size, subset = "")

nnSub(data, subset)

Arguments

formula	The formula to be used by nnet.
data	The dataframe to be used in the estimation.
decay	The decay parameter to be used by nnet.
size	The number of nodes in the hidden layer.
subset	A subseting expression (given as a quoted character string) for the estimation data frame.

Value

A set of components identical to those returned by nnet.

Author(s)

Dan Putler

See Also

nnet

Examples

  data(iris3)
  irisDat <- data.frame(rbind(iris3[,,1], iris3[,,2], iris3[,,3]), 
    species = as.factor(c(rep("s",50), rep("c",50), rep("v",50))))
  ir.nn2 <- Nnet(species ~ ., irisDat, 0.2, 2)

RcmdrPlugin.BCA documentation built on May 30, 2017, 7:57 a.m.