Register a response head

A response head says what the values being predicted are, how they reach a learner, and which cells of the (variable, fold) grid a score is defined on. Presence-absence with a joint multi-label head is what ships; an abundance or phenology response is a registration rather than a second fitting path.

Usage

register_response(name, spec, overwrite = FALSE)

responses()

Arguments

name

Name the response is asked for by.

spec

A list with elements prepare(y), returning the numeric matrix a learner is fitted on; activation, the name of the output transform ("sigmoid" or "identity"); loss, the name of the training objective ("binary_cross_entropy" or "squared_error"); metric, the default metric name; and cells(y, folds), returning the mask of scorable cells. Every learner that ships reads loss and activation from here: the encoders train under the loss and predict through the activation, and the learners fitting one model per response take the family the loss names, logistic or Gaussian. The combiner minimises the same loss. An optional weights(y) returns a [unit, variable] matrix of case weights every learner fits under; the shipped head's is positive_weights(), and a head without one fits unweighted.

overwrite

Replace an existing registration.

Value

The registered specification, invisibly.

Examples

responses()