Suppose you have implemented a classifier to predict whether…

Suppose you have implemented a classifier to predict whether a hospital patient has a certain condition. You would like to use the classifier’s results to develop a screening strategy for when additional medical testing (which can be costly) should be conducted on a patient. Using the estimated benefit of a true positive (value of diagnosing condition with additional medical testing) and the estimated cost of a false positive (cost of unnecessary medical testing), you construct a profit curve on a test set. Describe how to use the profit curve to establish whether additional medical testing should be conducted for a future patient.    

Consider a data set of 1000 observations to be used to train…

Consider a data set of 1000 observations to be used to train and validate a supervised learning algorithm. Compare how these data would be used in a 10-fold cross-validation procedure versus a static holdout procedure in which 70% of the data is used to train the model and 30% of the data is used to validate the model. For both of these approaches, compare: The number of different supervised learning models trained.  The size of the training set for each supervised learning model trained.  The total number of out-of-sample predictions on which performance measures are computed.    

A patient who is being admitted to the emergency department…

A patient who is being admitted to the emergency department with intermittent chest pain gives the following list of daily medications to the nurse.  Which medication has the most immediate implications for the patient’s care considering the medications the nurse will likely administer?