1' Compute the two measures sensitivity: and specicity: :t' .:-+3 " ' Repeat the last two steps for probability thresholds ranging from o to 1 in small steps {such as o.o1}. l Plot the pairs of sensitivity {on the yaxis} and 1specilfivity {itaids} on a scatterplot, and connect the points. The result is a curve called an RUE Curve. 1' The area under the RDC curve is called the AUC. Comput- ing this area is typically done using an algorithm. High AUC values indicate better performance, 1with v.5o inv dicating random performance and 1 denoting perfect perfor mance. {a} Using the logistic regression model that you tted in the last section, compute sensitivity and Ispecicity on the valr idation period for 'lE' following thresholds: o, v.1, v.2, n.3, o4, v.5, v.15, or}, v.3, n.9, 1. This can be easily done by mod ifying the probability threshold on the Excel i..R_Dutput worksheet; {b} Create a scatter plot of the 11 pairs and connect them. This is the RUE curve for your model. . Wl'iile AUC is a popular performance measure in competi- tions, it has been criticized for not being practically useful and even being flawed. In particular, Rice {solo} points out that in practice, a single probability threshold is typically used, rather than a range of thresholds. Other issues relate to lack of external validity and low precision. He suggests:"r