plotModels.ROC {FRESA.CAD}

Plot test ROC curves of each cross-validation model

Description

This function plots test ROC curves of each model found in the cross validation process. It will also aggregate the models into a single prediction performance, plotting the resulting ROC curve (models coherence). Furthermore, it will plot the mean sensitivity for a given set of specificities.

Usage

   plotModels.ROC(modelPredictions,
    number.of.models=0,
    specificities=c(0.975,0.95,0.90,0.80,0.70,0.60,0.50,0.40,0.30,0.20,0.10,0.05),
    theCVfolds=1,
    predictor="Prediction",
	cex=1.0,
	thr=NULL,
    ...)

Arguments

modelPredictions

A data frame returned by the crossValidationFeatureSelection_Bin function, either the Models.testPrediction, the FullBSWiMS.testPrediction,
the Models.CVtestPredictions, the TestRetrained.blindPredictions,
the KNN.testPrediction, or the LASSO.testPredictions value

number.of.models

The maximum number of models to plot

specificities

Vector containing the specificities at which the ROC sensitivities will be calculated

theCVfolds

The number of folds performed in a Cross-validation experiment

predictor

The name of the column to be plotted

cex

Controlling the font size of the text inside the plots

thr

The threshold for confusion matrix

...

Additional parameters for the roc function (pROC package)

Value

ROC.AUCs

A vector with the AUC of each ROC

mean.sensitivities

A vector with the mean sensitivity at the specificities given by specificities

model.sensitivities

A matrix where each row represents the sensitivity at the specificity given by specificities for a different ROC

specificities

The specificities used to calculate the sensitivities

senAUC

The AUC of the ROC curve that resulted from using mean.sensitivities

predictionTable

The confusion matrix between the outcome and the ensemble prediction

ensemblePrediction

The ensemble (median prediction) of the repeated predictions

Author(s)

Jose G. Tamez-Pena and Antonio Martinez-Torteya