multiDA

the multiDA function

Summarizes the trained multiDA classifier in a nice manner. User can select number of features to summarise

multiDA(X, y, penalty = c("EBIC", "BIC"), equal.var = TRUE,
  set.options = c("exhaustive", "onevsrest", "onevsall", "ordinal",
  "user"), sUser = NULL)

# S3 method for multiDA
plot(x, ranked = TRUE, ranks = 1:10,
  features = NULL)

# S3 method for multiDA
predict(object, newdata, ...)

# S3 method for multiDA
print(x, max.rank = 10, ...)

Arguments

X	matrix containing the training data. The rows are the sample observations, and the columns are the features.
y	vector of class values (for training)
penalty	default is in the form of the EBIC, which penalises based on the number of features. If option penalty="BIC" is specified, the penalty reverts back to the BIC.
equal.var	a LOGICAL value, indicating whether group specific variances should be equal or allowed to vary.
set.options	options for set partition matrix S.
sUser	if set.options is set to "user", sUser is a user input matrix for paritions to be considered. sUser MUST be a subset of the full partition matrix..
x	trained multiDA object
ranks	a vector of which ranked features should be plot
object	trained multiDA object
newdata	matrix of observations to predict. Each row corresponds to a new observation.
...	Any other variables which will be ignored
max.rank	number of significant features to display. If "ALL", all features are displayed.
x	object to print
...	Any other variables which will be ignored.

Value

multiDA object that contains the trained multiDA classifier

plots

list predicted class memberships of each row in newdata

Examples

#train the multiDA classifier using the SRBCT dataset, and find the resubstitution error rate

y   <- SRBCT$y
X   <- SRBCT$X
res  <- multiDA(X, y, equal.var=TRUE, set.options="exhaustive", penalty="EBIC")
vals <- predict(res, newdata=X)$y.pred          #y.pred returns class labels
rser <- sum(vals!=y)/length(y)