What question did this study set out to answer?

To compare classification rules that incorporate additional domain information against shrinkage-based rules and establish reliable estimators for their true error rates.

January 1, 2013Statistical Applications in Genetics and Molecular BiologyOpen Access

Performance and estimation of the true error rate of classification rules built with additional information. An application to a cancer trial

View Full Paper

Ask AI

Bookmark

Authors

DCDavid CondeUniversidad de Valladolid BSBonifacio SalvadorUniversidad de Valladolid CRCristina RuedaUniversidad de Valladolid

Discussion

Member takes

Implication

Methodological study reveals improved accuracy and error rate estimation for classification rules with additional information in cancer trial data, highlighting enhanced diagnostic reliability.

Key Points

To compare classification rules that incorporate additional domain information against shrinkage-based rules and establish reliable estimators for their true error rates.
Evaluated classification performance across four criteria: total misclassification probability, area under the ROC curve, well-calibratedness, and refinement.
Developed four new error rate estimators using leave-one-out bootstrap and cross-validation after bootstrap incorporating additional information.
Compared estimators against standard methods through Monte Carlo simulation studies and an empirical application to cancer clinical trial data.
Proved that the apparent error rate in rules using additional information requires specialized correction, demonstrating the need for tailored estimators.
Leave-one-out bootstrap estimators that integrate additional information consistently outperformed standard estimators in assessing the true error rate in both simulations and cancer trial data.

Cite This Study

Conde et al. (2013) studied this question.

synapsesocial.com/papers/6a1e8dac20458ccdef560c5a https://doi.org/10.1515/sagmb-2012-0037

View Full Paper

Ask AI

Bookmark

Implication

Methodological study reveals improved accuracy and error rate estimation for classification rules with additional information in cancer trial data, highlighting enhanced diagnostic reliability.

Performance and estimation of the true error rate of classification rules built with additional information. An application to a cancer trial

Discussion

Member takes

Implication

Key Points

Cite This Study

Performance and estimation of the true error rate of classification rules built with additional information. An application to a cancer trial

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider