Bayes' Law and Coin Flips

155 Views Asked by At

In Mitzenmacher's Probability and Computing there is the example of:

You are presented three coins. One of them is biased and will show heads with probability 2/3, the other two are fair, and show heads probability 1/2. Say that all three are flipped and the out come is HHT.

  1. First what is the probability that the first coins is biased, call this $E_1$? The second coin is biased, call this $E_2$? The third coin is biased, call this $E_3$? The answers are:

    $Pr(HHT | E_1) = \frac{2}{3} \cdot \frac{1}{2} \cdot \frac{1}{2} = \frac{1}{6}$

    $Pr(HHT | E_2) = \frac{1}{2} \cdot \frac{2}{3} \cdot \frac{1}{2} = \frac{1}{6}$

    $Pr(HHT | E_3) = \frac{1}{2} \cdot \frac{1}{2} \cdot \frac{1}{3} = \frac{1}{12}$

    Applying Bayes' Law we get:

    $Pr(E_1 | HHT) = \frac{Pr(HHT | E_1)Pr(E_1)}{\sum_{i=1}^{3} Pr(HHT | E_i)Pr(E_i)} = \frac{2}{5}$

    $Pr(E_2 | HHT) = \frac{2}{5}$

    $Pr(E_3 | HHT) = \frac{1}{5}$

  2. My question is, what if we had another coin flip permutation of HTH. So the question is something like, what is probability of $E_1$ given HTH AND given HHT? How would I solve this?