roc_ci: Compute Bootstrap Confidence Limits for Sensitivities and...

In sgutreuter/screenr: Construction of Binary Test-Screening Rules

Compute Bootstrap Confidence Limits for Sensitivities and Specificities

Description

roc_ci computes bootstrap confidence intervals from objects of class roc, as produced by the pROC package. roc_ci is simply a convenience wrapper for pROC::ci.thresholds re-formatted for screenr.

Usage

roc_ci(
  object,
  ...,
  bootreps = 4000,
  conf_level = 0.95,
  progress = "none",
  thresholds = "local maximas",
  se_min = 0.8
)

Arguments

object	an object of class roc.
...	additional arguments passed to pROC::ci.thresholds
bootreps	number of bootstrap replicates. Default: 4000.
conf_level	confidence level for uncertainty intervals. Default: 0.95.
progress	character-valued type of progress display (see help(pROC::ci.thresholds)). Default ⁠"none"⁠.
thresholds	type of thresholds (see help(pROC::ci.thresholds)).
se_min	minimum value of sensitivity returned. Default: 0.8.

Value

roc_ci returns a dataframe containing thresholds with their sensititives, specificities and uncertainy intervals.

See Also

ci.thresholds

sgutreuter/screenr documentation built on Oct. 19, 2024, 12:49 p.m.