Create the settings for a fusion ensemble — createFusionCombiner • EnsemblePatientLevelPrediction

Create the settings for a fusion ensemble

createFusionCombiner(
  type = c("uniform", "AUROC", "AUPRC")[1],
  evaluation = "CV",
  scaleFunction = "normalize"
)

Arguments

type: The type of fusion ensemble pick from: 'uniform', 'AUROC', 'AUPRC' or any other metric from evaluationSummary
evaluation: The evaluation type used to learn the weights (if evaluation is CV and type is 'AUROC' then the cross validation AUROC is used to determine the weight given to of the base models)
scaleFunction: How to scale the weights (normalize means weights add up to 1)

Examples

if (FALSE) {

comSettingNormAUC <- createFusionCombiner(type = "AUROC",
                                          evaluation = "CV",
                                          scaleFunction = "normalize")

comSettingUniform <- createFusionCombiner(type = "uniform",
                                          evaluation = "CV",
                                          scaleFunction = "normalize")

}