Markov's inequality, unclear Wikipedia intuition

153 Views Asked by At

Search for the word "intuitive" here. I do not understand how

$$E[X]=0\cdot \bar{a}+\frac{E[X]}{a}\cdot a$$

shows anything at all. I would like to see an intuitive proof of an upper bound of $P(x\geq a)$ which should be $$P(x\geq a)\leq \frac{E[X]}{a}.$$

1

There are 1 best solutions below

0
On

For an increasing $\varphi$, $X\geq a \implies \varphi(X)\geq\varphi(a)$. Thus you can construct a similar argument. However, remember it is important to have $X$ a non-negative random variable in the original argument and thus if we are to replicate the argument the image of $\varphi$ should be non-negative.

Note that what you request, i.e.

$$P(|X|\geq a)\leq \frac{E[\varphi(X)]}{\varphi(a)}$$

can not be true in general. Take $X\sim N(0,1)$, $a=1$ and $\varphi$ the identity.