grnn.test {BETS}

Test a set of General Regression Neural Networks

Description

Given new values of the independent variables, tests a list of trained GRNNs and picks the best net, based on an accuracy measure between the forecasted and the actual values.

Usage

grnn.test(results, test.set)

Arguments

results

The object returned by grnn.train.

test.set

A ts list. The first element must be the actual values of the dependent variable. The others, the new values of the regressors.

Value

A list object representing the best network (according to forecasting MAPE). Its fields are:

• mape: The forecasting MAPE

• model: The network object

• sigma: The sigma parameter

• id: The id number of the network, as given by grnn.train

• mean: The predicted values

• x: The original series

• fitted: The fitted values

• actual: The actual values (to be compared with the predicted values)

• residuals: Difference between the fitted values and the series original values

• regressors: The regressors used to train the network

Author(s)

Talitha Speranza [e-mail]