You are interested in training a binary classifier for a data set consisting of 100 observations of three continuous inputs and a binary outcome. You will consider three different models: logistic regression with additive terms, logistic regression with all pair-wise interactions, and a neural network. You will assess the performance of the three models using 5-fold crossvalidation. You are interested in identifying the model with the highest area under the ROC curve. What threshold value is used to create the ROC curve? Depends on the application 0.5 0.25 or 0.75 Many thresholds between 0 and 1