โ€œํ™•๋ฅ ๊ณผ ํ†ต๊ณ„(MATH230)โ€ ์ˆ˜์—…์—์„œ ๋ฐฐ์šด ๊ฒƒ๊ณผ ๊ณต๋ถ€ํ•œ ๊ฒƒ์„ ์ •๋ฆฌํ•œ ํฌ์ŠคํŠธ์ž…๋‹ˆ๋‹ค. ์ „์ฒด ํฌ์ŠคํŠธ๋Š” Probability and Statistics์—์„œ ํ™•์ธํ•˜์‹ค ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค ๐ŸŽฒ

2 minute read

โ€œํ™•๋ฅ ๊ณผ ํ†ต๊ณ„(MATH230)โ€ ์ˆ˜์—…์—์„œ ๋ฐฐ์šด ๊ฒƒ๊ณผ ๊ณต๋ถ€ํ•œ ๊ฒƒ์„ ์ •๋ฆฌํ•œ ํฌ์ŠคํŠธ์ž…๋‹ˆ๋‹ค. ์ „์ฒด ํฌ์ŠคํŠธ๋Š” Probability and Statistics์—์„œ ํ™•์ธํ•˜์‹ค ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค ๐ŸŽฒ

<์ฒด๋น„์‡ผํ”„์˜ ๋ถ€๋“ฑ์‹; Chebyshevโ€™s Inequality>์€ ํ‰๊ท  $\mu$๋กœ๋ถ€ํ„ฐ $\lambda$ ๊ฑฐ๋ฆฌ ์ด์ƒ ๋ฉ€์–ด์ง„ ๊ฒฝ์šฐ, ์ฆ‰ tail ์ƒํ™ฉ์— ๋Œ€ํ•œ ํ™•๋ฅ ์˜ ์ƒํ•œ์„ ์ œ์‹œํ•œ๋‹ค. ์ฆ‰, โ€œThe upper bound of tail probabilityโ€์ธ ์…ˆ์ด๋‹ค. ์‹์€ ์•„๋ž˜์™€ ๊ฐ™์ด ์ •์˜๋˜์–ด ์žˆ๋‹ค.

Theorem. Chebyshevโ€™s Theorem

Let $X$ be a RV with $\text{Var}(X) < \infty$ and let $\lambda > 0$, then

\[P \left( \left| X - \mu \right| \ge \lambda \right) \le \frac{\text{Var}(X)}{\lambda^2}\]

์‚ฌ์‹ค <Chebyshevโ€™s inequality>๋Š” ํ‰๊ท ์œผ๋กœ๋ถ€ํ„ฐ ๋ฐ”๊นฅ์ชฝ๋ณด๋‹ค๋Š” ํ‰๊ท  ์•ˆ์ชฝ์— ๋Œ€ํ•œ ํ™•๋ฅ ์„ ๊ตฌํ•  ๋•Œ ์ฃผ๋กœ ์‚ฌ์šฉํ•œ๋‹ค.

Example.

Supp. a RV $X$ has $\mu = 8$ and $\sigma^2 = 9$. Show that $P(0 < X < 16) \ge \dfrac{55}{64}$.

Sol.

\[\begin{aligned} P(0 < X < 16) &= P(-8 < X -\mu <8) \\ &= 1 - P(\left| X - \mu \right| \ge 8) \\ &\ge 1 - \frac{\sigma^2}{8^2} = 1 - \frac{9}{64} = \frac{55}{64} \end{aligned}\]


<Chebyshevโ€™s Theorem>์˜ ์ฆ๋ช…์€ ์ƒ๊ฐ๋ณด๋‹จ ๊ฐ„๋‹จํ•˜๋‹ค.

Proof.

\[P \left( \left| X - \mu \right| \ge \lambda \right) = \int_{\{ x \; : \; \left| x - \mu \right| \ge \lambda \}} 1 \cdot f(x) dx \le \int^{\infty}_{-\infty} 1 \cdot f(x) dx\]

์ด๋•Œ, $P \left( \left| X - \mu \right| \ge \lambda \right)$์—์„œ $\left| X - \mu \right| \ge \lambda$๋ผ๋Š” ์กฐ๊ฑด์ด ์žˆ์œผ๋ฏ€๋กœ

\[\left| X - \mu \right| \ge \lambda \iff \left| \frac{X - \mu}{\lambda} \right| \ge 1 \iff \left( \frac{X - \mu}{\lambda} \right)^2 \ge 1\]

๋”ฐ๋ผ์„œ ์ด๋ฅผ ์œ„์˜ ์ ๋ถ„์‹์— ์ ์šฉํ•˜๋ฉด,

\[\begin{aligned} P \left( \left| X - \mu \right| \ge \lambda \right) &\le \int^{\infty}_{-\infty} 1 \cdot f(x) dx \\ &\le \int^{\infty}_{-\infty} \left( \frac{X - \mu}{\lambda} \right)^2 \cdot f(x) dx \\ &= \frac{\text{Var}(X)}{\lambda^2} \end{aligned}\]

$\blacksquare$


<์ฒด๋น„์‡ผํ”„ ๋ถ€๋“ฑ์‹>์€ ์ดํ›„์— ํ†ต๊ณ„(Statistics) ํŒŒํŠธ์—์„œ <Weak Law of Large Numbers>๋ฅผ ์ฆ๋ช…ํ•  ๋•Œ, ํ™œ์šฉํ•œ๋‹ค. ์ž์„ธํ•œ ๋‚ด์šฉ์€ ์•„๋ž˜์˜ ํฌ์ŠคํŠธ๋กœ ๊ณ ๊ณ ~

๐Ÿ‘‰ Sampleing Distribution of Mean: Weak Law of Large Numbers