Neural Networks Using Model Averaging
Aggregate several neural network models
"avNNet"(x, y, repeats = 5, bag = FALSE, allowParallel = TRUE, seeds = sample.int(1e+05, repeats), ...) "avNNet"(formula, data, weights, ..., repeats = 5, bag = FALSE, allowParallel = TRUE, seeds = sample.int(1e+05, repeats), subset, na.action, contrasts = NULL)"predict"(object, newdata, type = c("raw", "class", "prob"), ...)
A formula of the form
class ~ x1 + x2 + ...
matrix or data frame of
xvalues for examples.
- matrix or data frame of target values for examples.
- (case) weights for each example -- if missing defaults to 1.
- the number of neural networks with different random number seeds
- a logical for bagging for each repeat
- random number seeds that can be set prior to bagging (if done) and network creation. This helps maintain reproducibility when models are run in parallel.
- if a parallel backend is loaded and available, should the function use it?
Data frame from which variables specified in
formulaare preferentially to be taken.
- An index vector specifying the cases to be used in the training sample. (NOTE: If given, this argument must be named.)
A function to specify the action to be taken if
NAs are found. The default action is for the procedure to fail. An alternative is
na.omit, which leads to rejection of cases with missing values on any required variable. (NOTE: If given, this argument must be named.)
- a list of contrasts to be used for some or all of the factors appearing as variables in the model formula.
an object of class
avNNetas returned by
- matrix or data frame of test examples. A vector is considered to be a row vector comprising a single case.
Type of output, either:
rawfor the raw outputs,
codefor the predicted class or
probfor the class probabilities.
arguments passed to
Following Ripley (1996), the same neural network model is fit using different random number seeds. All the resulting models are used for prediction. For regression, the output from each network are averaged. For classification, the model scores are first averaged, then translated to predicted classes. Bagging can also be used to create the models.
If a parallel backend is registered, the foreach package is used to train the networks in parallel.
- a list of the models generated from
- an echo of the model input
- if any predictors had only one distinct value, this is a character string of the remaining columns. Otherwise a value of
avNNet, an object of
"avNNet.formula". Items of interest in the output are:
Ripley, B. D. (1996) Pattern Recognition and Neural Networks. Cambridge.
data(BloodBrain) ## Not run: # modelFit <- avNNet(bbbDescr, logBBB, size = 5, linout = TRUE, trace = FALSE) # modelFit # # predict(modelFit, bbbDescr) # ## End(Not run)