SAS代写: MTH 491B – Final Project

MTH 491B – Final Project

For this project, you will be designing a product for your insurance company. One example of a product is All State’s Accident Forgiveness Plan. As you can see from our model (namely in the Parameters Excel sheet), if a person gets into even one accident, their rates go up. Essentially what All State’s plan does is if the person gets into one and only one accident, the rates do not change. There could be many ways to modify your product.

 For a person involved in only one accident, treat their accident charge as if it were the rate for zero accidents.

o If he or she gets into a second accident, treat this rate as a 2-accident rate  Essentially, this is removing the 1-accident rate from the table.

o If he or she gets into multiple accidents, treat the number of accidents as n – 1 accidents.

 Essentially, this is just bumping the table down one unit. For questions related to this:

 How much does this change the premium? o Higher / more profitable?

o Lower / less profitable? o Speculation

  •   Do you think this would bring in more or less customers?
  •   A lower premium means more customers, but does the increase in business

    equate to an increase in profit?

  •   Consider that more drivers means more potential accidents. If you advertise for

    accident forgiveness, will your new insureds likely be better or worse drivers?

    Project Outline
    Your project will consist of designing two products for the company. You may not use the Accident Forgiveness product. You may even analyze the data when you change the rates (you may adjust mortality assumptions, the smoker/loyalty/accident discounts, etc., again, with explanation). If you decide to change rates, no matter how many you change, this only does not count as a product. This may help you adjust your product to recoup some of the loss of premium or help lower high premiums after issuing the product. Analyze each product independently and then together. Analysis can consist of measures like we have done before:

  •   Aggregate premium
  •   Marginal premiums (such as premiums by ZIP code)
  •   Distributions
  •   Others

    The goal here is to make a projection on what you are suggesting as a product. It is alright to make assumptions, but you want to see what those assumptions imply, and explain why they may or may not be acceptable assumptions as above. The project is open ended and all you’re trying to do is come up with a creative product and have a meaningful discussion about it that is backed up by facts, when available.

Keep in mind that the data set does not have any loss variable. The annual premium listed is only income to the company, so if you restrict business, your company will take a loss no matter what, by design. You may want to make a somewhat arbitrary loss/benefit variable so you can compare premium income to benefit payment. Don’t make it too arbitrary, though – come up with a somewhat reasonable model and even cite where you found your model if you do research on modeling. This also does not count as a product.

The bottom line is: No matter what you decide to do, be prepared to explain and defend your work!

What You Will Be Turning In
This will be a slideshow presentation. I would suggest avoid putting code in the slideshow unless it is a very integral portion of explaining your model. Submit your slideshow and your code to the D2L dropbox.

Good luck!