Why we need symmetry when we apply Chebyshev's inequality?

163 Views Asked by At

I have a question about the given solution of this problem.

Problem: A farmer uses $1000$ gallons of water on average per day. The variance of the water rate does not exceed $40000$ gallons. What's the probability that on any given day the water rate will not exceed $2000$ gallons? Use Chebyshev's inequality.

Given solution: Let $X$ be the daily water rate in gallons.

Because the interval $0 \le X \le 2000$ is symmetrical in terms of $EX=1000$, we can apply Chebyshev's inequality $$P\left(X\le2000\right)=P\left(0\le X\le2000\right)=P\left(\left|X-1000\right|\le1000\right)\ge1-\frac{40000}{1000^{2}}=0.96$$

My questions:

  1. Why do we need symmetry here to apply Chebyshev's inequality?
  2. What if we need to find the probability that on any given day the water rate will not exceed $1500$ or $2500$ gallons?

I appreciate any help.

1

There are 1 best solutions below

3
On BEST ANSWER

Why do we need symmetry here to apply Chebyshev's inequality?

Chebishev inequality gives an approximated bound for the probability of the following event

$$|X-\mu|<\epsilon$$

that is

$$\mu-\epsilon\leq X\leq \mu+\epsilon$$

which is a symmetric interval w.r.t. the mean

Are you sure they want you to calculate $P(X<1500)$ using Chebischev inequality? I think only $P(X<2500)$ can be calculated using this inequality.


EDIT: if you want to calculate $P(X<1500)$ given that the mean is 1000 it is clear that Chebishev inequality can give an approx of the probability that

$$|X-1000|<500$$

that is

$$500<X<1500$$

and this is not what you want...

On the contrary, if you are interested in $P(X<2500)$ you can set

$$|X-1000|<1500$$ that is

$$-500<X<2500$$

and this is good for you as it is the same as

$$0<X<2500$$

given that $X$ cannot be negative.