Data Assignment | Homework For You
Question 1. We are given data that has been aggregated over a few years of MBA admission records of a university. The program is designed to run for two years. The passing grade in the program is a B, which has a numeric score of 3.0. Students with a GPA of less than 3.0 at the end of two years cannot graduate even if they complete their coursework. Therefore students with a GPA of less than 3.0 at the end of the first are put on probation.
In this data set, we have the undergraduate GPA and several test scores at the time of admission, and the GPA of the students at the end year1. Our goal is to build a model to predicts students who will be on probation at the end of year1. The MAB Office thinks that such a model would provide an early warning signal so that such students are followed up closely and advised.
This model can be attempted in several different ways. You are asked to turn in one model. Briefly outline why and how you ended up with this model. Provide a discussion of its predictive performance with a detailed discussion of the way you define and evalute performance in this case.
Suppose your model is used to issue early warnings signals so that such students receive close monitoring and support. What could be anticipated consequences??
Question 2. Consider the car insurance problem from the perspective of a car owner in Turkey and only within the scope of accident risk. Damages due to earthquakes, floods, etc are not part of the discussion for simplification purposes. There are mainly two types of insurance policies that cover accident damages: Liability insurance which is mandatory and collision damage insurance which is optional. Liability insurance would cover the damges that you are libale for. In other words, If a driver is involved in an accident and is at fault, the insurance would cover the damages suffered by other parties but not the driver’s own daamge. In this case, If the driver has collision damage insurance, damages to self woulr be covered too.
a) Formulate a particular car owner’s collision damage insurance purchase decision by using a decision tree. Populate your tree with realistic monetary outcomes by assuming a particular car. The book value of a car and its insurance cost canbe obtained usingg online brokerage sites. You will have to model/ estiamte accident probabilities and costs.
b) Solve teh descision tree using EMV criterion. Is the collision damage insurance purchase decision justified?
c) Consider an alternative policy with deductible amount of %5 of the car’s value and with an insurance Premium of %40 of the one you used aove. With such a dedctible policy, a car owner would pay for damages up to %5 of the car’s value Ä°f an accident happens and insurance would cover the rest. How does thic compare to the policy above fort he particular case you studied above? In general, a policy with a dedutible would appeal to what kind of customers?
d) Would building a simulation model have any advantages over a decision tree model for this decision problem? You do not need to build a simulation model. Abrief discussion is sufficient. Get Statistics homework help today