Detalhes bibliográficos
| Resumo: | The ROC curve is a statistical tool used broadly to help professionals from several fields of study gauge the ability of a binary classifier. Recent theoretical advancements have allowed the ROC curve to better examine existing confounding variables in its analysis allowing greater calibration for markers and classifiers. A few packages developed for the R language have already incorporated these newfound concepts and are currently available to aid users in the covariate study. This article combines different ROC curve, adjusted ROC curve and covariate specific ROC curve methodologies across packages to study the effect of sex on the CRIB score system with a resampling strategy using parallel computing. Results show a confounding effect on roughly 15% of cases with similar results across packages confirming a consensus among methods and providing a robust methodology for future use. |
|---|
| Autores principais: | Machado e Costa, Francisco |
|---|
| Outros Autores: | Braga, A. C. |
|---|
| Assunto: | AROC Resampling ROC curve |
|---|
| Ano: | 2021 |
|---|
| País: | Portugal |
|---|
| Tipo de documento: | comunicação em conferência |
|---|
| Tipo de acesso: | acesso restrito |
|---|
| Instituição associada: | Universidade do Minho |
|---|
| Idioma: | inglês |
|---|
| Origem: | RepositóriUM - Universidade do Minho |
|---|