Fall 2015 Computer Science 458 10/5/15

Announcements

• export DISPLAY - another try.

• Here is a draft list of sample final projects.

• Final projects are due the last day of reading period.

• Office hours. Ronghui Gu is holding office hours at 305 AKW.

  Wednesday 2-4 pm.
  Thursday 7-9 pm.
  

  Kun Run has office hours at 203 AWK.

  Monday 7-9 pm
  Friday 10am - 12pm
  

• Ronghui has created a class Piazza site for the course. You should have received an email on friday notifying you of your enrollment. This should facilitate communication. https://piazza.com/

• hw2: sensitivity function. Here is how it should work: You pass it a list of options, just as with decide. sensitivity then simulates ten different decisions by varying one of the five numeric assumptions either up or down. It applied a given variation to each option, and then runs decide on that new option pair. It might make sense for sensitivity first to call decide with no variation, to establish a base case.

  So you might have the following series of calls within sensitivity:

  1. decide(optOH, OptSC)
  2. decide(optOHlowcost, OptSClowcost)
  3. decide(optOHhighcost, OptSChighcost)
  4. decide(optOHlowdiscount, OptSClowdiscount)
  5. decide(optOHlowcostpercar, OptSClowcostpercar)
  6. usw. (etc.)

  This is a first level approach. You could add more sophistication by calculating the correlations among the variations. You could also include the probabilities of each scenario and then calculate an expected value for all possibilities, if that makes sense.

  Finally, you need to be prepared to take handle other options, besides the ones specified. For example, you could have two Ohio plant options, one of which is the given one, and another with a lower cost factory that produces fewer cars.

• Next Monday's guest speaker is Professor Andrew Lo from MIT. There will be dinner at Mory's afterwards as well.

  Here some of his recent papers. He will discuss one or more of them. You are welcome to ask him questions about any of them.

  • Privacy-Preserving Methods for Sharing Financial Risk Exposures
  • Consumer credit-risk models via machine-learning algorithms
  • Moore’s Law versus Murphy’s Law: Algorithmic Trading and Its Discontents
  • Risk and Risk Management in the Credit Card Industry

• Next Wednesday's guest speakers are from Facebook. Here is a paper that is related to one of the topics they will discuss.
  • Pairwise choice as a simple and robust method for inferring ranking data

Mid Term Exam

• There will be a mid term exam in class on Monday October 19th. I understand that CS 323 has a midterm scheduled that day as well. I will offer a make up exam, no deans excuse required. It is your choice.

• The exam will mostly comprise programming questions in R and Python. Any finance questions will be framed as programming questions. For example, here is the formula for computing the beta of a stock portfolio. Implement that in R or Python.

• The scope of the questions will be as follows:
  • R: The R Cookbook, chapters 1 - 9.
  • R: stockPortfolio package.
  • R: the swirl tutorial.
  • Python: The Python Cookbook, chapters 1 - 8.

• The questions will primarily be short answer, e.g., what output does this code produce? or what code would produce this output?

Discussion

• See R package: stockPortfolio

• Example code

• Modern Portfolio Theory Continue with Efficient Market Hypothesis.

  Answers to exercies have been posted.

• Case Based Systems