ROC632: Construction of diagnostic or prognostic scoring system and internal validation of its discriminative capacities based on ROC curve and 0.633+ boostrap resampling.

This package computes traditional ROC curves and time-dependent ROC curves using the cross-validation, the 0.632 and the 0.632+ estimators. The 0.632+ estimator of time-dependent ROC curve is useful to estimate the predictive accuracy of prognostic signature based on high-dimensional data. For instance, it outperforms the other approaches, especially the cross-validation solution which is often used. The 0.632+ estimators correct the area under the curve in order to adequately estimate the prognostic capacities regardless of the overfitting level. This package also allows for the construction of diagnostic or prognostic scoring systems (penalized regressions). The methodology is adapted to complete data (penalized logistic regression associated with ROC curve) or incomplete time-to-event data (penalized Cox model associated with time-dependent ROC curve).

Install the latest version of this package by entering the following in R:
install.packages("ROC632")
AuthorY. Foucher <Yohann.Foucher@univ-nantes.fr>
Date of publication2013-12-27 22:22:51
MaintainerY. Foucher <Yohann.Foucher@univ-nantes.fr>
LicenseGPL (>= 2)
Version0.6
www.r-project.org
www.divat.fr

View on CRAN

Files

Questions? Problems? Suggestions? or email at ian@mutexlabs.com.

Please suggest features or report bugs with the GitHub issue tracker.

All documentation is copyright its authors; we didn't write any of that.