Confidence bands

Calculate confidence bands for a single timepoint timeROC ROC based on threshold averaging as described in Fawcett 2004.

confidencebands(
  T,
  delta,
  marker,
  timepoint,
  cause = 1,
  ...,
  nboot = 200,
  conf.level = 0.95,
  verbose = interactive()
)

Arguments

T

numeric, time, see timeROC::timeROC().

delta

numeric, status, see timeROC::timeROC().

marker

numeric, marker, see timeROC::timeROC().

timepoint

numeric(1), timepoint to evaluate.

cause

numeric, cause, see timeROC::timeROC().

nboot

numeric(1), number of boostrap samples.

conf.level

numeric(1), confidence level.

verbose

logical, if TRUE a progressbar is shown.

...

further arguments passed to timeROC.

Value

a list with 2 elements (TP, FP), each one a matrix, with 3 rows (lower/median/upper) band, and length(threshold) columns.

References

Fawcett, T. (2004). ROC graphs: Notes and practical considerations for researchers. Machine learning, 31(1), 1-38.