Ok...But in ROC curve a N*2 matrix is the input. In case of determining the ROC for PLS-DA model what should be these two columns? I can generate a ROC curve with other softwares like statistica or medcal. But I am unable to decide my input variables.
Hi Jianguo..I suppose you have developed the web based statistical analysis tool for metabolomics..Can you please help me interpret one of the permutation test statistics data. If you give your email I can send it to you.
There are many softwares and web tools you can use to produce your ROC. The key point is what to use as input to asses the PLS-DA. You could try using your PLS-DA score on the discrimination direction to test. Better results are usually obtained with O-PLS-DA and the tp1 score of the horizontal axis.