WeightedCombinationComputer: WeightedCombinationComputer

In frbl/OnlineSuperLearner: Online SuperLearner package

Description

WeightedCombinationComputer

Usage

WeightedCombinationComputer

Format

An object of class R6ClassGenerator of length 24.

Methods

initialize(weights.initial)

Creates a new computer for determining the best weighted combination of the ML libraries. weigths.initial vector containing the initial weights. @param weights.initial vector the initial vector of weights to use.

compute(Z, Y, libraryNames, ...)

Method to compute the best weighted combination in an underlying optimizer. Note that by default this method is not implemented, and should be implemented by one of the subclasses.

@param Z matrix containing the outcomes of each of the estimators.

@param Y vector containing the actual observed outcomes.

@param libraryNames vector containing the names of the estimators.

@param ... other parameters to pass to the underlying combination computers.

process(Z, Y, libraryNames, ...)

Method to compute the best weighted combination of the underlying optimizer. This function internally calls the extended compute function.

@param Z matrix containing the outcomes of each of the estimators.

@param Y vector containing the actual observed outcome.

@param libraryNames vector containing the names of the estimators.

@param ... other parameters to pass to the underlying combination computers.

@return vector of the trained / updated weights.

get_weights

Active Method. Returns the current vector of optimal weights (or the initial weights, if not yet fitted)

@return vector the current vector of weights

get_historical_weights()

Active Method. Returns the history of the alphas, and shows how they were updated over time.

@return data.frame a data.frame containing the alphas. One row for each iteration, one column for each algorithm.

get_step_count

Active Method. Returns the current step count of the algorithm. This is updated on every process call.

@return integer the current stepcount

frbl/OnlineSuperLearner documentation built on Feb. 9, 2020, 9:28 p.m.