2) For each of the following scenarios, state which type of machine learning would be appropriate: classification, regression, clustering, or a Pattern discovery. Briefly (1-2 sentences) justify your answers.

a) You are having an argument with your friend about how many social groups there are at your school. You believe there are about half-a-dozen natural groups based on tastes in things like music, clothes, athletics, and politics, while your friend thinks everyones tastes are random. You discover you can access (publicly available) individual records from a poll where 1000 students scored their preferences on 20 forms of arts and entertainment.

b) You work at an oil company, and they are interested in predicting whether wells drilled in several new formations will produce oil or not. They give you a large quantity of data from past drilling efforts (geographic location, depth of well, type of rock, age of formation, etc.), along with the success or failure of each drilled well.

c) The florist in your neighborhood has a pretty good idea what kinds of flowers arrangements her steady customers like, but shed like to be more scientific and send personalized coupons for her customers. She has excellent records of all their past purchases, as well as complete histories of which arrangements they have viewed on the shops website.

d) A dietician has been trying to understand how peoples dietary choices affect the amount of weight they gain or lose but isnt seeing obvious patterns. For a recent 6-month period, he has good records for 150 of his clients on their consumption of 12 different foods, along with the change in their weight over that period.

e) A coffee shop manager in UW Bothell started to collect transactions record to find information that can help predict whether a customer is likely to be lost to a competitor coffee shop on Campus.