In the opinion of the instructor, Logistic Regression models will be the most widely used type of analytics that you will encounter. Think of an application of logistic regression, discuss the benefits of the model, and discuss the difficulties of implementing such a model.

EXAMPLE (and you can't use this one because I just took it!):

http://www.easternwinelabs.com/

In the wine industry, there is interest in knowing ahead of time if a particular vintage will result in good bottles of wine or bad bottles. There is no way to know if the wine is good until it is finished aging. This might take several years. So researchers in the wine industry collected environmental data about the climate, soil, and fertilizers that were available during the grapes growing season. This data is used as inputs into a model. The output is whether the wine is considered good or bad. Thus this is a boolean model. The model can be used to predict which wine should be purchased in bulk, based upon the information.

Although the benefits of this are obvious, there are three obvious difficulties. First, it is difficult to collect the data. Climate, soil, and fertilizer information are from different sources and must be collected and merged into a data set. Second, the target data is highly subjective. Who gets to decide if a wine is "good" or "bad". Two wine experts might disagree. Lastly, the target is not necessarily truly a binary target. People tend to grade a wine on a scale from 1 to 100. So a boolean model, although simple, might have issues with consistency because different people might like the wine or dislike the wine. Also, the intensity of their opinion might be different.