In: Statistics and Probability
Suppose there are 1000 urns in a mathematician's warehouse. Each of these urns contains 20 marbles. 900 of the urns contain 15 green and 5 blue marbles, while 100 contain only blue marbles. Call these Type I and Type II urns, respectively. Suppose that we randomly select an urn and randomly draw a marble form it.
A. What is the probability that we selected a Type I urn, given that the ball we draw is blue? Given that the ball we draw is green?
B. Suppose I only catch a partial glimpse of the ball when it is drawn. This leads me to have 80% confidence that a blue ball was drawn. Assuming I update by Jeffrey Conditionalization, how confident should I then be that the ball was drawn from a Type I urn?
Given:
1000 urns -> Type I (900) with 15 green and 5 blue balls each | Type II (100) with 20 blue balls each
A) Given that a blue ball is drawn, probability that Type I urn is selected:- P(Type I urn|blue ball)
Using Bayes Theorem, we know that,
So, using the above theorem,
Now, P(Blue Ball) Can be written as (using total probability theorem
Thus,
Similarly,
Thus,
We calculate P(Type I| Greenball) using the same method as above
P(Type I/Greenball) = = 1
1 or 100% probability is obvious here, as green is only present in Type I
B)
We know that P(Blue ball) = 0.8, P(Greenball) = 0.2
Using Jeffrey Conditionalization, we can say that
P(Type I) = P(Type I|Blueball) . P(Blueball) + P(Type I|Greenball) . P(Greenball)
=
=0.754
I am 75.4% confident that ball was drawn from Type I