DATA 352/551: Assignment 1

Group discussion based on Homework 1

Group activity

Work in groups of 2-3. Your group will submit one response on this Google doc. Make sure to put all group members name on your response. We will discuss every group’s response in class.

Part 1: Exact probability versus simulation

  1. Let \(X\) be the number of heads in three tosses of a fair coin. Imagine simulating 20, 100, and 10000 sets of three tosses. What would you expect to happen to the simulated proportions as the number of repetitions increases?

  2. A group reports the following results from 20 repetitions:

    Number of heads 0 1 2 3
    Simulated count 1 8 7 4

    Which features of these results are consistent with the theoretical pmf? What would you avoid concluding from only 20 repetitions?

Part 2: Design a fair comparison

Your group wants to compare the probability that three fair dice have a sum of 9 with the probability that they have a sum of 10. One student proposes simulating until the first sum of 9 appears and then comparing the number of rolls with the number of rolls until the first sum of 10 appears. Discuss why this is less useful than recording both outcomes across the same fixed number of repetitions.

Part 3: The martingale and misleading summaries

In the roulette martingale, a player stops after winning at least 5 dollars or losing more than 100 dollars. Red occurs with probability \(18/38\).

Discuss the following claim:

“The strategy is safe because most simulated games end with a win of 5 dollars.”

As a group, respond to these questions:

  1. What additional quantities would help us understand the risk besides the proportion of games that end in a win?
  2. How could a small number of large losses affect the average net winnings, the median net winnings, and the proportion of winning games differently?
  3. Write a two- or three-sentence conclusion that would be more informative than the claim above.

Part 4: Why do smaller hospitals vary more?

In one town, a large hospital has about 45 births per day and a small hospital has about 15 births per day. The probability of a boy is 0.5 at both hospitals. Why can the small hospital have more days above 60% even though the underlying probability of a boy is the same?