Monotone Spline Risk Estimates

Fits a Monotone constrained Generalized Additive Model (GAM) to estimate risk, given a vector of binary outcomes and a vector of scores.

Usage

getMSPLINEest(outcome, score, k = 10, fitonPerc = TRUE)

Arguments

outcome

Vector of binary outcome for each observation.

score

Numeric vector of continuous predicted risk score.

k

Numeric to specify the upper limit of basis functions to fit for GAM. See mgcv::s() for more details. Defaults to -1.

fitonPerc

Logical; if TRUE, fit gam on risk percentiles. Defaults to TRUE.

Value

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

See also

getASISest() getBINNEDest() getCGAMest() getGAMest() getPAVAest()

Examples

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

tail(getMSPLINEest(outcome = truth, score = rscore), 10)
#>         score percentile outcome   estimate
#> 324 0.3472267 0.81081081       0 0.39218604
#> 325 0.2940563 0.63063063       0 0.35211614
#> 326 0.2956584 0.63963964       0 0.35636133
#> 327 0.3173334 0.70870871       0 0.38029566
#> 328 0.1957009 0.21621622       0 0.08332628
#> 329 0.2909602 0.61561562       0 0.34466754
#> 330 0.1466197 0.03603604       0 0.08332628
#> 331 0.2335213 0.40240240       0 0.18794378
#> 332 0.1310011 0.01201201       0 0.08332628
#> 333 0.3696760 0.85885886       0 0.41581142