I'm having an issue with a question regarding Bayes' Theorem. Here is the question:

An online clothing store carries three brands of jeans. 40% of sales are brand A, 20% are brand B and the remainder are brand C. 20% of brand A jeans cost over 100, 40% of brand B jeans cost over $100 and 90% of brand C jeans cost over 100. Given that a pair of jeans is purchased for over 100, what is the probability that they are brand A?

My work is as follows:

Using Bayes Theorem I classified two events

$A$: Picking brand A

$B$: Jeans cost over 100

So I would need to find $P(A|B)$

$$P(A) = 2/5$$

$$P(B|A) = 1/5$$

$$P(\neg A) = 3/5$$

$$P(B|\neg A) = 11/25$$

When I do calculations with Bayes' theorem, I get;

$$\frac{2/25}{2/25 + (11/25)\times(3/5)} = 10/43$$

But the answer is $2/13$. Now, when I first did the question, I forgot to multiply $P(B|\neg A)$ by $P(\neg A)$ in the denominator, and I got the right answer. Is there a reason I should leave $P(\neg A)$ out? Or did I approach the problem in the wrong way completely?

  • $\begingroup$ Where do you get $\mathrm{P}(\mathsf{B}|!\mathsf{A})$ from ? (And does this need a 'homework' tag?) $\endgroup$ – Scortchi Dec 6 '12 at 1:04
  • 2
    $\begingroup$ I would suggest that you give some thought to your choice of notation. Given that A, B, and C are already used as name brands and you have decided to use A to mean Brand A is chosen, you are just setting yourself for confusion by choosing B to denote Jeans cost over \$100 instead of B being chosen. Couldn't you have thought of something else? Say O for over \$100 the way that both answers did? $\endgroup$ – Dilip Sarwate Dec 6 '12 at 3:33

This would be correct:


Not this:



It is easier to say that you have 3 disjoint events:

$A$: picking up brand A

$B$: picking up brand B

$C$: picking up brand C

and you want to compute $P(A|O)$ with

$O$: jeans cost over 100


$$P(A|O) = \frac{P(O|A)P(A)}{P(O)} = \frac{P(O|A)P(A)}{P(O|A)P(A) + P(O|B)P(B) + P(O|C)P(C)} = \\ \frac{20/100 \cdot 40/100}{20/100 \cdot 40/100 + 40/100 \cdot 20/100 + 90/100 \cdot 40/100} = 8/52 = 2/13$$

In your case $P(B|\neg A)$ cannot be computed by $40/100 \cdot 20/100 + 90/100 \cdot 40/100 = 11/25$.

  • $\begingroup$ That seems like a more clear cut way to do the question. Thank-you. $\endgroup$ – user1209379 Dec 6 '12 at 2:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.