predReg: Registry Ensemble Prediction

In VizOR: Graphical Visualization Tools for Complex Observational Data with Focus on Health Sciences

Description

Generate an rms:Predict object or data frame for an ensemble of simulated disease registries

Usage

predReg(genReg, N, M = 100, fit = NULL,
  adjust.to = fit$Design$limits["Adjust to", ],
  do.pred = function(df) {    
    fit.call <- fit$call
    fit.call$data <- quote(df)
    fit <- eval(fit.call) 
    fit$Design$limits["Adjust to", names(adjust.to)] <- adjust.to
    if(is(fit, "lrm"))
      Predict(fit, fun = plogis)
    else
      Predict(fit) },
  ...)

Arguments

genReg	A function that returns a simulated registry dataset, taking as its first parameter the desired size of the simulated registry, and possibly other parameters passed through via the ... arg
N	Size of generated registries
M	Size of the ensemble
fit	A fitted model usually intended to serve as a template for a model to be fitted to the simulated registries. This may be NULL when do.pred is provided explicitly in the call
adjust.to	A list of adjust-to values for the fitted models, defaulting to the adjust-to parameters of fit
do.pred	A function to be run on the data generated during each iteration of the simulation, generating either an rms:Predict object, or a (usually, named) atomic vector
...	Additional parameters passed to genReg

Details

Given a function for generating a simulated disease registry, this function generates an ensemble of such registries. It then returns an rms:Predict object that contains ensemble-averaged predictions and confidence bounds.

Value

Depending on the return type of do.pred, either an rms:Predict object containing ensemble-averaged predictions with confidence bounds reflecting their estimated ensemble variance, or else a data frame collecting the vector returned by do.pred

Author(s)

David C. Norris

VizOR documentation built on May 30, 2017, 5:29 a.m.