Let $X_1,X_2,...,X_n$ be a fixed number of Bernoulli random variables. My problem is to find a distribution for $Y$ such that for some function $f$, we have $Y=f(X_1,X_2,...,X_n)$. There are two candidate functions to use, $max$ or $avg$. I have no idea if an average function would work here or not but I think it'd give me a meaningful result.

I have looked into similar problems and mostly found the cases where $X_i$ are continuous random variables. Any hint on this problem is highly appreciated.


  • $\begingroup$ This sounds like homework. Also, it's not clear what you mean by two candidate functions. Both max and avg, along with dozens of other functions f will give you a distribution. Are there conditions on Y that you've forgotten to include? $\endgroup$ – Gschneider Apr 5 '12 at 13:01
  • 1
    $\begingroup$ It sounds like you're just being asked to derive the distribution of the average of a bunch of bernoulli trials (think about a scaled binomial distribution) and the maximum. There are a number of threads about deriving the distribution of the maximum of a set of random variables, so I'd suggest searching for those $\endgroup$ – Macro Apr 5 '12 at 13:07
  • 2
    $\begingroup$ @user10405 Those were polite and justified comments; there is no need to react emotionally. $\endgroup$ – user88 Apr 6 '12 at 7:45

User10405, I didn't mean to offend you by the homework comment, but proving that the sum of independent Bernoullis is a Binomial is exactly a homework problem I assigned the statistics class (Mathematical Statistics 1) I teach recently. Apologies for the "useless prejudice."

I'm confused, Macro's comment seemed appropriate, yet:

Macro: This is not a bunch of Bernoulli trials and a binomial distribution does not apply.

But in the original statement:

Let $X_1,X_2,...,X_n$ be a fixed number of Bernoulli random variables

See http://en.wikipedia.org/wiki/Bernoulli_random_variable for verification that the sum of Bernoullis is indeed a Binomial, and since $avg$ is just the sum divided by $n$, it will be a scaled Binomial.

As far as $f=max$ goes, since each of the X 's is a Bernoulli, i.e. 0 or 1, the max also has to be 0 or 1. It will be 0 only if all of the X 's are 0. So treat it as a Bernoulli($p∗$) where $p∗=1−P(X_1=X_2=...=X_n=0)$ .

An assumption of independence among the $X$'s makes $p∗$ much easier to calculate.

I get the feeling that this answer still won't be what you're looking for, but without more clarification, I don't see how we can help further.

  • 1
    $\begingroup$ Yes this is how I was thinking of the problem. If the candidate function is the average, then $Y$ having a (scaled) binomial distribution would have the desired property and you've given such a distribution for $max$. Either a crucial detail is missing from the problem description, or our point is not being fully understood. $\endgroup$ – Macro Apr 6 '12 at 1:35

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.