doMulti(object, top, method, ...)
"doMulti"(object, top, method, ...)
ExprsArray
object to undergo feature selection.top = 0
to include all features. A numeric vector can also be used
to indicate specific features by location, similar to a character vector.ExprsBinary
method to execute multiple times.method
.
ExprsMulti
: Method to execute multiple "1 vs. all" binary tasks.
doMulti
depends on the total number of levels in the
$defineCase
factor. If a training set is missing any
one of the factor levels (e.g., owing to random cuts during
cross-validation), the ExprsModule
component that
would refer to that class label gets replaced with an NA
placeholder. This NA placeholder gets handled as a
special case when predicting with an ExprsModule
.During ExprsModule
class prediction, the absence
of a class during training (i.e., an NA placeholder)
will prevent an ExprsModule
object from possibly
predicting that class in a validation set. Rather, an
ExprsModule
can only make predictions about class
labels that it "knows". However, all "unknown" classes
in the validation set (i.e., those missing from the training
set) still impact metrics of classifier performance.
An ExprsModule
object can only make predictions on
an ExprsMulti
object with the same number of recorded
class labels (i.e., the total number of levels in the
$defineCase
factor). As with all functions included
in this package, all ties get broken using probability
weights proportional to the relative class frequencies
in the training set.
fs
build
doMulti
exprso-predict
plCV
plGrid
plGridMulti
plMonteCarlo
plNested