Statistical Analysis of ROC Curves
StAR is a server that computes ROC curves and several related statistics to assess the significance of their differences in performance.
StAR can receive as input several classifiers, compute their corresponding ROC curves and display the results graphically. The area under the curve (AUC) and other values obtained at the optimal threshold are also displayed.
The core of the server relies on a non-parametric test for the difference of the AUC that accounts for the correlation of the ROC curves. This test takes advantage of the equality between the Mann-Whitney U-statistic for comparing distributions and the AUC when computed by the trapezoidal rule. The server is primarily designed for paired data. However, it can also analyse balanced unpaired data when the number of units is the same for each classifier. It cannot be used with partially-paired data.
For more details about the statistical method implemented here, please read the following manuscript:
Vergara, I.A., Norambuena, T., Ferrada, E., Slater, A.W. and Melo, F. (2008) StAR: a simple tool for the statistical comparison of ROC curves. BMC Bioinformatics 9, 265.
If you use this server in your work and find it useful, please acknowledge it just by citing the reference above.