I have a random variable $X$, whose c.d.f. is known. Now I want to estimate the c.d.f of $Y = X \cdot Z$, where the only information I have about $Z$ is that it is a discrete r.v. which takes on three values.

What is the best estimate(in MMSE sense??) for the c.d.f. of $Y$.

My intuition says that it must be the c.d.f. of $X$, but I cant show it rigorously?

  • $\begingroup$ With no further information about the distribution of $Z$, this is undoable. $\endgroup$ – Did Jan 29 '12 at 17:30

Suppose $Z$ assumes one of three possible values $z_1$, $z_2$ or $z_3$ with certain probabilities: $$ \mathbb{P}(Z = z_k ) = p_k \quad k=1,2,3 $$ such that $p_1 + p_2 + p_3 = 1$. Since $X$ and $Z$ are independent, $$ F_Y(y) = \mathbb{P}(Y \leqslant y) = \mathbb{P}(z_1 X \leqslant y) p_1 + \mathbb{P}(z_2 X \leqslant y) p_2 + \mathbb{P}(z_3 X \leqslant y) p_3 $$ If we further assume that $z_1 > 0$ and $z_2>0$ and $z_3>0$ we get: $$ F_Y(y) = p_1 F_X\left(\frac{y}{z_1}\right) + p_2 F_X\left(\frac{y}{z_2}\right) + p_3 F_X\left(\frac{y}{z_3}\right) $$


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.