GridsearchBreaks: Gridsearch breaks function

In citronmeliss/predictionpackr: Predictionpackr

Usage

GridsearchBreaks(predictions, outcome, grid = seq(0.01, 0.99, 0.01),
  n.breaks = 3, loss.function = "auc", sample = TRUE,
  parallel = FALSE, n.cores = NULL, save.plot = FALSE,
  verbose = FALSE, maximise = TRUE)

Arguments

predictions	Numeric vector. Predicted probabilites. No default.
outcome	Numeric vecotr. The outcome of interest. No default.
grid	Numeric vector. All values to be combined and searched. Defaults to seq(0.01, 0.99, 0.01).
n.breaks	Numeric vector of length 1. The number of breaks in the predicted probabilites, i.e. cut-points. Defaults to 3.
loss.function	Character vector of length 1. The name of the loss metric, e.g. auc. For now any performance measure included in ROCR is accepted. Defaults to "auc".
sample	Logical. If TRUE loss is only calculated for 10 \item parallelLogical. If TRUE the gridsearch is done in parallel. Defaults to FALSE. \item n.coresNumeric vector of length 1. The number of cores to use for parallel computing. Defaults to NULL. \item save.plotLogical. If TRUE a wireframe plot is saved to disk. Defaults to FALSE. \item verboseLogical. If TRUE then information is printed as the gridsearch runs. Defaults to FALSE. Currently ignored. \item maximiseLogical. If TRUE, grid search maximizes performance metric. Defaults to TRUE.

Performs a gridsearch to identify breaks to optimise a provided loss function

citronmeliss/predictionpackr documentation built on Feb. 10, 2020, 12:19 a.m.