0
$\begingroup$

I am calculating Customer Lifetime Value and have the following expression:

Customer Lifetime Value = Exp( visits | lifetime) * avg(transaction size)

In context of statistics, what does Exp(z | y) mean?

What is Exp in this context? (exponential, e^(z | y)..?)

What is | in this context? (z*y, z given y..?)

Thanks.

$\endgroup$
4
  • $\begingroup$ CLV probably is a random variable, so on the right side of equation you should have a 'recipe' for random variable, and avg is a real number. That being said, Exp(visits|lifetime) could be a conditional expectation. See wikipedia for conditional expectation. $\endgroup$ Feb 5, 2017 at 11:05
  • $\begingroup$ Thank you. So E(z | y) is more correct? $\endgroup$ Feb 5, 2017 at 11:10
  • 1
    $\begingroup$ I don't know how about you, but I'm used to writing random variables with capitals, so it would be rather E(Z|Y). Small letters 'z' and 'y' usually mean observed value of random variable Z and Y, respectively. $\endgroup$ Feb 5, 2017 at 11:34
  • $\begingroup$ There are two different things: for a real integrale random variable $X$ and another random variable there is the conditional expectation $E[X|Y]$, see en.m.wikipedia.org/wiki/Conditional_expectation. This is yet another random variable. Then there is the factorization of that, denoted by $E[X|Y=y]$, that is, a function from the space Y maps to to the reals. This quantity is so important in ML because given features X and a real valued outcome Y, one can show that y->E[X|Y=y] is the best regression (with lowest RMSE). They mean E[Visits|Lifetime=lifetime]. $\endgroup$ Jun 27, 2019 at 6:13

1 Answer 1

1
$\begingroup$

Yes, in this case Exp is meant to be $\mathit{E}$, the expectation of number of visits given the lifetime of customer.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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