Dear sir/madam, I subjected the data obtained from MEDME package to ROCR analysis as suggested using these commands: pred <- prediction(predictions, labels) perf <- performance(pred, measure = "tpr", x.measure = "fpr") plot(perf, col=rainbow(10)) A brief set of "prediction" values in the data ($predictions): 0.59836348 0.42820548 0.88141522 0.07484118 0.41030317 0.34349624 0.63349003 0.51225317 0.31789225 0.74818798 0.02281738 0.36549982 0.40980097 0.25466337 0.27680128 0.20879282 0.26144171 0.59784408 1.16341073 0.64634278 0.02603932 0.74096199 0.38492218 0.03429591 0.15660236 1.04699480 0.04065977 0.95849380 1.24050240 0.03268187 0.17247822 1.19127392 1.02523731 0.04234674 0.25061179 0.45305541 However ,based on what criteria do you assign the binary values- $labels: (1,0)?and also, how do you correlate these label values to the values of the prediction set? Thanking you, Yours Sincerely, Susan Bosco, MSc in Molecular Biology and Human Genetics, Manipal Life Sciences Centre, Manipal-India. The INTERNET now has a personality. YOURS! See your Yahoo! Homepage. [[alternative HTML version deleted]]