PAVA Risk Estimates

Determines isotonic regression estimates via pava, given a vector of binary outcomes, and a vector of scores.

Usage

getPAVAest(
  outcome,
  score,
  weights = rep(1, length(outcome)),
  ties = "primary",
  low_events = NULL,
  low_nonevents = NULL,
  high_events = NULL,
  high_nonevents = NULL,
  hilo_obs = NULL
)

Arguments

outcome

Vector of binary outcome for each observation.

score

Numeric vector of continuous predicted risk score.

weights

Vector of numerics to specify PAVA observation weighting.

ties

String to specify how ties should be handled for PAVA.

low_events

Numeric, specifying number of events in the lowest bin.

low_nonevents

Numeric, specifying number of nonevents in the lowest bin.

high_events

Numeric, specifying number of events in the highest bin.

high_nonevents

Numeric, specifying number of nonevents in the highest bin.

hilo_obs

Numeric, specifying number of observations in the highest and lowest bins.

Value

A data frame with 4 columns (score, score percentile, outcome, estimate).

See also

getASISest() getBINNEDest() getCGAMest() getGAMest() getMSPLINEest()

Examples

# Read in example data
auroc <- read.csv(system.file("extdata", "sample.csv", package = "stats4phc"))
rscore <- auroc$predicted
truth <- as.numeric(auroc$actual)

tail(getPAVAest(outcome = truth, score = rscore), 10)
#>         score percentile outcome   estimate
#> 324 0.3472267 0.81081081       0 0.41818182
#> 325 0.2940563 0.63063063       0 0.32203390
#> 326 0.2956584 0.63963964       0 0.32203390
#> 327 0.3173334 0.70870871       0 0.38095238
#> 328 0.1957009 0.21621622       0 0.08602151
#> 329 0.2909602 0.61561562       0 0.32203390
#> 330 0.1466197 0.03603604       0 0.08602151
#> 331 0.2335213 0.40240240       0 0.09803922
#> 332 0.1310011 0.01201201       0 0.08602151
#> 333 0.3696760 0.85885886       0 0.41818182