Joe is a new analyst in your organization. He is anxious to apply all of the new modeling techniques he learned in his Masters program. A majority of your company's business comes from repeat buyers. As a result you have created a program to reach out to customers who have not made a purchase in the last six months. It is imperative that you keep those customer active as customers who go 8 months or more without a purchase are 90% likely to never make a purchase again. Anxious to add value, Joe creates a file from your customer base and runs all of the records through a neural network. The file included all customers and had a y variable which was coded 1 (made a purchase last 6 months) and 0 (did not make a purchase last 6 months) and several predictor variables (including age, gender, buying habits.) The resulting Neural Network had a missclassification rate of 18%. Based on these findings, Joe is ready to use the model to predict. Based only on this description, which of the following statements is not true?

Question 9Select one:

a.

Given the information provided, we do not know whether or not a Neural Network was the best choice to model this particular problem.

b.

Without understanding the base rate of occurrence (what percent made a purchase in the last 6 months), we do not know whether the missclassification rate is a good measure of model performance.

c.

There seems to be no validation or test set, so it is likely that the model overfit the build sample.

d.

Given the nature of the problem, a Neural Net is not an appropriate technique in this situation.