Two Cultures

Reading Questions

  1. What are the two cultures outlined by Breiman?
  2. Is the Ozone Project supervised or unsupervised? Classification or Regression? Which methods that we’ve seen could be used to tackle this problem?
  3. What is the name of the model/method that is discussed in equation \(R\) of section 5.1?
  4. In section 5.4 he states, “If the model has too many parameters, then it may overfit the data and give a biased estimate of accuracy”. Where would this model be in terms of the bias-variance tradeoff?
  5. What is the Rashoman effect? Did you run into this effect is question 5 from the last lab?
  6. Explain how one of the techniques that we’ve covered could be seen to invoke Occam’s Razor.

Discussion Questions

  1. The most illuminating point for me in this paper was…
  2. The most confusing point for me in this paper was…
  3. Which of the responses (Cox, Efron, Hoadley, Parzen) do you find the most incisive? Why?
  4. Which do you think is the strongest single criticism of Breiman’s paper that is levelled by the commentators?
  5. The big ticket question: in your area of study, if you had to use methods from only one of Breiman’s cultures for the rest of your life, which would it be: Data Model or Algorithmic Model?