calPlot2: Calibration plots for binary data
In ModelGood: Validation of risk prediction models

Description Usage Arguments Details Value Author(s) References Examples

Calibration plots for risk prediction models in for a binary endpoint

calPlot2(object, formula, data, splitMethod = "none", B = 1, M, showY,
  method = "nne", round = TRUE, bandwidth = NULL, q = 10,
  density = 55, add = FALSE, diag = !add, legend = !add, axes = !add,
  xlim, ylim, xlab = "Predicted event probability",
  ylab = "Observed proportion", col, lwd, lty, pch, cause = 1,
  percent = TRUE, giveToModel = NULL, na.action = na.fail, cores = 1,
  verbose = FALSE, ...)

`object`	A named list of prediction models, where allowed entries are (1) R-objects for which a predictStatusProb method exists (see details), (2) a `call` that evaluates to such an R-object (see examples), (3) a matrix with predicted probabilities having as many rows as `data` in one column. For cross-validation all objects in this list must include their `call`.
`formula`	A survival or event history formula. The left hand side is used to compute the expected event status. If `formula` is `missing`, try to extract a formula from the first element in object.
`data`	A data frame in which to validate the prediction models and to fit the censoring model. If `data` is missing, try to extract a data set from the first element in object.
`splitMethod`	Defines the internal validation design: `none/noPlan`: Assess the models in the give `data`, usually either in the same data where they are fitted, or in independent test data. `BootCv`: Bootstrap cross validation. The prediction models are trained on `B` bootstrap samples, that are either drawn with replacement of the same size as the original data or without replacement from `data` of the size `M`. The models are assessed in the observations that are NOT in the bootstrap sample.
`B`	The number of cross-validation steps.
`M`	The size of the subsamples for cross-validation.
`showY`	If `TRUE` the observed data are shown as dots on the plot.
`method`	The method for estimating the calibration curve(s): `"nne"`: The expected event status is obtained in the nearest neighborhood around the predicted event probabilities. `"quantile"`: The expected event status is obtained in groups defined by quantiles of the predicted event probabilities.
`round`	If `TRUE` predicted probabilities are rounded to two digits before smoothing. This may have a considerable effect on computing efficiency in large data sets.
`bandwidth`	The bandwidth for `method="nne"`
`q`	The number of quantiles for `method="quantile"`.
`density`	Gray scale for observations.
`add`	If `TRUE` the line(s) are added to an existing plot.
`diag`	If `FALSE` no diagonal line is drawn.
`legend`	If `FALSE` no legend is drawn.
`axes`	If `FALSE` no axes are drawn.
`xlim`	Limits of x-axis.
`ylim`	Limits of y-axis.
`xlab`	Label for y-axis.
`ylab`	Label for x-axis.
`col`	Vector with colors, one for each element of object. Passed to `lines`.
`lwd`	Vector with line widths, one for each element of object. Passed to `lines`.
`lty`	lwd Vector with line style, one for each element of object. Passed to `lines`.
`pch`	Passed to `points`.
`cause`	For competing risks models, the cause of failure or event of interest
`percent`	If TRUE axes labels are multiplied by 100 and thus interpretable on a percent scale.
`giveToModel`	List of with exactly one entry for each entry in `object`. Each entry names parts of the value of the fitted models that should be extracted and added to the value.
`na.action`	Passed to `model.frame`
`cores`	Number of cores for parallel computing. Passed as the value of the argument `mc.cores` when calling `mclapply`.
`verbose`	if `TRUE` report details of the progress, e.g. count the steps in cross-validation.
`...`	Used to control the subroutines: plot, axis, lines, legend. See `SmartControl`.

For method "nne" the optimal bandwidth with respect to is obtained with the function dpik from the package KernSmooth for a box kernel function.

list with elements: time, Frame and bandwidth (NULL for method quantile).

Thomas Alexander Gerds

TA Gerds, PA Andersen, and Kattan MW. Calibration plots for risk prediction models in the presence of competing risks. Statistics in Medicine, page to appear, 2014.

set.seed(40)
N=40
Y=rbinom(N,1,.5)
X1=rnorm(N)
X1[Y==1]=rnorm(sum(Y==1),mean=rbinom(sum(Y==1),1,.5))
X2=rnorm(N)
X2[Y==0]=rnorm(sum(Y==0),mean=rbinom(sum(Y==0),3,.5))
dat <- data.frame(Y=Y,X1=X1,X2=X2)
lm1 <- glm(Y~X1,data=dat,family="binomial")
lm2 <- glm(Y~X2,data=dat,family="binomial")
calPlot2(list(lm1,lm2),data=dat)