β€œν™•λ₯ κ³Ό 톡계(MATH230)” μˆ˜μ—…μ—μ„œ 배운 것과 κ³΅λΆ€ν•œ 것을 μ •λ¦¬ν•œ ν¬μŠ€νŠΈμž…λ‹ˆλ‹€. 전체 ν¬μŠ€νŠΈλŠ” Probability and Statisticsμ—μ„œ ν™•μΈν•˜μ‹€ 수 μžˆμŠ΅λ‹ˆλ‹€ 🎲

7 minute read

β€œν™•λ₯ κ³Ό 톡계(MATH230)” μˆ˜μ—…μ—μ„œ 배운 것과 κ³΅λΆ€ν•œ 것을 μ •λ¦¬ν•œ ν¬μŠ€νŠΈμž…λ‹ˆλ‹€. 전체 ν¬μŠ€νŠΈλŠ” Probability and Statisticsμ—μ„œ ν™•μΈν•˜μ‹€ 수 μžˆμŠ΅λ‹ˆλ‹€ 🎲

λ“€μ–΄κ°€λ©°

μ–΄λ–€ 사건이 ν‰κ· μ μœΌλ‘œ β€œ5뢄에 1λ²ˆβ€ λ°œμƒν•œλ‹€κ³  κ°€μ •ν•΄λ³΄μž. 예λ₯Ό λ“€μ–΄, μΆœν‡΄κ·Ό μ‹œκ°„μ— 2ν˜Έμ„  μ§€ν•˜μ² μ΄ ν‰κ· μ μœΌλ‘œ 5λΆ„λ§ˆλ‹€ ν•œ λŒ€μ”© λ„μ°©ν•œλ‹€κ³  생각할 수 μžˆλ‹€.

λΈ”λ£¨ν˜Όμ€ μ‚¬λ‹Ήμ—­μ—μ„œ 2ν˜Έμ„  μ§€ν•˜μ² μ„ 기닀리고 μžˆμŠ΅λ‹ˆλ‹€. μˆ˜λ§Žμ€ μΆœν‡΄κ·Ό κ²½ν—˜μ— μ˜ν•΄ λΈ”λ£¨ν˜Όμ€ 이 μ‹œκ°„λŒ€μ— ν‰κ· μ μœΌλ‘œ 5λΆ„ 정도 기닀리면 μ§€ν•˜μ² μ΄ μ˜¨λ‹€λŠ” 것을 μ•Œκ³  μžˆμŠ΅λ‹ˆλ‹€. μ–΄λ–¨ λ•ŒλŠ” 2ν˜Έμ„ μ„ λˆˆμ•žμ—μ„œ 놓쳐도 λ‹€λ₯Έ λ‹€μŒ μ—΄μ°¨κ°€ λ“€μ–΄μ™€μ„œ 3뢄도 μ•ˆ 기닀릴 λ•Œκ°€ μžˆμ§€λ§Œ, μ–΄λ–¨ λ•ŒλŠ” 2ν˜Έμ„ μ„ ν•˜μ—Όμ—†μ΄ 기닀릴 λ•Œλ„ μžˆμŠ΅λ‹ˆλ‹€.

평균적인 λŒ€κΈ° μ‹œκ°„μ„ μ•Œκ³  μžˆμ„ λ•Œ, μ§€ν•˜μ² μ„ 기닀리기 μœ„ν•΄ μ“°λŠ” μ‹œκ°„μ€ β€œμ§€μˆ˜ λΆ„ν¬β€λΌλŠ” 연속 ν™•λ₯  뢄포λ₯Ό λ”°λ¦…λ‹ˆλ‹€!

Distribution for waiting time

λ‹€μŒμ— λ“€μ–΄μ˜¬ 2ν˜Έμ„ μ„ κΈ°λ‹€λ¦¬λŠ”λ° κ±Έλ¦¬λŠ” μ‹œκ°„μ€ μ•„λž˜μ˜ μ§€μˆ˜ 뢄포λ₯Ό λ”°λ¦…λ‹ˆλ‹€.

Definition. Exponential Distribution (waiting time)

Let $\beta >0$ is an average waiting time, and we say that $X$ has an <exponential distribution> with parameter $\beta$ if it has pdf $f(x)$ as

\[f(x) = \begin{cases} \frac{1}{\beta} e^{- \frac{1}{\beta} x} & \text{for} \; x > 0\\ \quad 0 & \text{else} \end{cases}\]

and we denote such RV $X$ as $X \sim \text{EXP}(\beta)$.

Expectation and Variance

$\beta$λ₯Ό 평균 λŒ€κΈ° μ‹œκ°„(average waiting time)으둜 μ •μ˜ν–ˆμœΌλ―€λ‘œ, ν™•λ₯  λ³€μˆ˜ $X$의 κΈ°λŒ“κ°’(즉, 평균)은 $\beta$와 κ°™μŠ΅λ‹ˆλ‹€. 즉,

\[E[X] = \beta\]

λΆ„μ‚°μ˜ κ²½μš°λŠ” μ§€μˆ˜ ν™•λ₯  뢄포에 λŒ€ν•œ 식을 잘 μ •λ¦¬ν•˜λ©΄ μ•„λž˜μ˜ κ²°κ³Όλ₯Ό 얻을 수 μžˆμŠ΅λ‹ˆλ‹€.

\[\text{Var}(X) = \beta^2\]

μ΄λ•Œ $\beta$λŠ” 사건이 λ°œμƒν•˜κΈ° μœ„ν•΄ ν‰κ· μ μœΌλ‘œ λŒ€κΈ° ν•˜λŠ” μ‹œκ°„μ„ 의미 ν•©λ‹ˆλ‹€. β€œ5뢄에 1건씩” λ°œμƒν•˜λŠ” 경우라면, $\beta = 5$κ°€ 되고, λΆ„ν¬λŠ” μ•„λž˜μ™€ κ°™μŠ΅λ‹ˆλ‹€.

\[P(X = x) = \frac{1}{5} e^{-\frac{1}{5} x}\]

Distribution for event rate

λΈ”ν˜Όμ€ μΆœν‡΄κ·Ό μ‹œκ°„μ— μ§€ν•˜μ² μ—­μ—μ„œ λ‹€μŒ μ—΄μ°¨λ₯Ό κΈ°λ‹€λ¦¬λŠ” λ™μ•ˆ, μ „κ΄‘νŒμ„ λ³΄λ©΄μ„œ β€œ1μ‹œκ°„ λ™μ•ˆ μ—΄μ°¨κ°€ λͺ‡ λŒ€λ‚˜ μ§€λ‚˜κ°”μ„κΉŒ?β€λΌλŠ” ꢁ금증이 μƒκ²ΌμŠ΅λ‹ˆλ‹€. λŒ€κΈ° μ‹œκ°„μ— λŒ€ν•œ 뢄포λ₯Ό μ‚΄νŽ΄λ΄€λ˜ κ²ƒμ²˜λŸΌ, μ§€ν•˜μ² μ΄ 평균 5λΆ„λŒ€ 1λŒ€μ”© λ„μ°©ν•œλ‹€κ³  κ°€μ •ν•˜λ©΄, 1μ‹œκ°„ λ™μ•ˆ λ„μ°©ν•˜λŠ” μ—΄μ°¨μ˜ κ°―μˆ˜λŠ” 12λŒ€κ°€ 될 것 μž…λ‹ˆλ‹€.
κ·ΈλŸ¬λ‚˜ μ–΄λ–€ 날은 μŠ€ν¬λ¦°λ„μ–΄ κ³ μž₯으둜 μ—΄μ°¨κ°€ 지연 λ˜μ–΄ 1μ‹œκ°„ λ™μ•ˆ 10λŒ€κ°€ 올 수 있고, μ–΄λ–€ 날은 μš΄ν–‰μ΄ μˆœμ‘°λ‘œμ›Œμ„œ 1μ‹œκ°„μ— 15λŒ€κ°€ 올 μˆ˜λ„ μžˆμŠ΅λ‹ˆλ‹€.

λΈ”λ£¨ν˜Όμ€ μΌμ •ν•œ μ‹œκ°„ λ™μ•ˆ μ—΄μ°¨κ°€ λͺ‡ λŒ€ λ„μ°©ν•˜λŠ”μ§€λ₯Ό ν™•λ₯ μ μœΌλ‘œ λͺ¨λΈλ§ν•˜κ³  μ‹Άμ–΄μ‘ŒμŠ΅λ‹ˆλ‹€. 사건 λ°œμƒ 횟수(count)에 λŒ€ν•œ ν™•λ₯  뢄포가 λ°”λ‘œ β€œν‘Έμ•„μ†‘ λΆ„ν¬β€μž…λ‹ˆλ‹€. 푸아솑 뢄포에 λŒ€ν•΄μ„œλŠ” 별도 ν¬μŠ€νŠΈμ— μ •λ¦¬ν•œ 것도 μžˆμŠ΅λ‹ˆλ‹€.

Definition. Poisson Distribution (event rate)

Let $\lambda >0$ is an event occurring rate, and we say that $X$ has an <Poisson distribution> with parameter $\lambda$ if it has pdf $f(x)$ as

\[f(x) = \frac{\lambda^x \cdot e^{-\lambda}}{x!}\]

for $x = 0, 1, …$ and we denote such RV $X$ as $X \sim \text{POI}(\lambda)$.

μ˜ˆμ‹œμ˜ 상황을 κ°€μ Έμ™€μ„œ ν•¨μˆ˜λ₯Ό λͺ¨λΈλ§ ν•˜λ©΄, 5λΆ„λ‹Ή 1λŒ€μ˜ μ—΄μ°¨κ°€ λ“€μ–΄μ˜¨λ‹€λ©΄ 1λΆ„λ‹Ή 0.2λŒ€μ˜ μ—΄μ°¨κ°€ λ“€μ–΄μ˜€λŠ” 것과 κ°™μŠ΅λ‹ˆλ‹€. 즉, $\lambda = 0.2$ 이것을 포아솑 λΆ„ν¬μ˜ ν•¨μˆ˜λ‘œ 적으면

\[P(X = x) = \frac{(0.2)^x \cdot e^{-0.2}}{x!}\]

와 κ°™μŠ΅λ‹ˆλ‹€.

Expectation and Variance

Theorem.

Let $X \sim \text{EXP}(\lambda)$, then

  • $E[X] = \dfrac{1}{\lambda}$
  • $\text{Var}(X) = \dfrac{1}{\lambda^2}$

뢄산에 λŒ€ν•œ κ²ƒλ§Œ

Proof.

$\lambda = \beta = 1$인 ν‘œμ€€ μ§€μˆ˜ 뢄포에 λŒ€ν•΄ μƒκ°ν•΄λ΄…μ‹œλ‹€.

\[Y \sim \text{EXP}(1)\]

이 ν‘œμ€€ μ§€μˆ˜ λΆ„ν¬μ˜ 평균과 뢄산은 μ–΄λ–»κ²Œ λ κΉŒμš”?

\[\begin{aligned} E[Y] &= \int^{\infty}_0 y \cdot e^{-y} \; dy = 1 \end{aligned}\]

Varianceλ₯Ό ꡬ해보면,

\[\begin{aligned} E[Y^2] = \int^{\infty}_0 y^2 \cdot e^{-y} \; dy = 2 \end{aligned}\]

λ”°λΌμ„œ, $\text{Var}(Y) = E[Y^2] - E[Y]^2 = 2 - 1^2 = 1$.


이제, $X \sim \text{EXP}(\beta = 1 / \lambda)$λ₯Ό μ‚΄νŽ΄ λ΄…μ‹œλ‹€. λŒ€κΈ° μ‹œκ°„μ΄ $\beta$ 만큼 λŠ˜μ–΄λ‚¬μœΌλ―€λ‘œ $X = \beta \cdot Y = \dfrac{Y}{\lambda}$λ₯Ό λ§Œμ‘±ν•©λ‹ˆλ‹€. λ”°λΌμ„œ

\[E[X] = E\left[\beta \cdot Y \right] = E\left[\frac{Y}{\lambda}\right] = \frac{1}{\lambda}\]

그리고 뢄산은

\[\text{Var}(X) = \text{Var}\left( \beta \cdot Y \right) = \text{Var}\left( \frac{Y}{\lambda} \right) = \frac{1}{\lambda^2}\]

Unit Conversion

If $X \sim \text{EXP}(1)$, then $Y := \dfrac{X}{\lambda} \sim \text{EXP}(\lambda)$.

\[P(Y > y) = P(\frac{X}{\lambda} > y) = P(X > \lambda y) = e^{-\lambda y}\]

본인은 μœ„μ˜ 상황을 (minute - second) λ³€ν™˜μ„ λ°”νƒ•μœΌλ‘œ μ΄ν•΄ν–ˆλ‹€. λ§Œμ•½ $X$κ°€ λΆ„ λ‹¨μœ„μ—μ„œ 처음 λ„μ°©ν•˜λŠ” λ²„μŠ€μ˜ μ‹œκ°„μ„ λͺ¨λΈλ§ν•˜κ³ , κ·Έ λ•Œμ˜ parameterκ°€ $\lambda = 1$라고 ν•˜μž. μš°λ¦¬λŠ” 이것을 초 λ‹¨μœ„μΈ $60X$둜 λ³€ν™˜ν•  수 μžˆλ‹€. μ΄λ•Œμ˜ tail probabilityλŠ”

\[P(60X > x) = P(X > x/60) = e^{- x/60}\]

λ”°λΌμ„œ, $60X \sim \text{EXP}(1/60)$이 λœλ‹€. 이것은 $60X$μ—μ„œ $\lambda$κ°€ $\lambda = 1/60$이 됨을 μ˜λ―Έν•œλ‹€. μ΄λ•Œ, $\lambda$λŠ” Poisson Process의 parameter둜, Time Unit λ‹Ή λ„μ°©ν•˜λŠ” λ²„μŠ€μ˜ 수λ₯Ό λͺ¨λΈλ§ν•œλ‹€. λ”°λΌμ„œ 1초 λ‹Ή ν‰κ· μ μœΌλ‘œ 1/60 λŒ€μ˜ λ²„μŠ€κ°€ 도착함을 μ˜λ―Έν•œλ‹€. 이것을 $\beta = 1 / \lambda$둜 ν•΄μ„ν•˜λ©΄, λ²„μŠ€κ°€ ν•œλ²ˆ λ„μ°©ν•˜λŠ” μ‹œκ°„μ΄ ν‰κ· μ μœΌλ‘œ 60μ΄ˆκ°€ 됨을 μ˜λ―Έν•œλ‹€!

더 λ‚˜μ•„κ°€κΈ°

Duality: Exponential Distribution and Poisson Process

푸아솑 κ³Όμ •μ—μ„œ 연속적인 두 사건 μ‚¬μ΄μ˜ 간격은 μ§€μˆ˜ 뢄포λ₯Ό λ”°λ¦…λ‹ˆλ‹€. λ‚΄μš©μ΄ κΈΈμ–΄μ Έμ„œ 별도 포슀트둜 λΆ„λ¦¬ν•˜μ˜€μŠ΅λ‹ˆλ‹€ γ…Žγ…Ž

πŸ‘‰ Duality: Exponential Distribution and Poisson Process

Duality: Exponential Distribution and Geometric Distribution

연속확λ₯ λΆ„포인 μ§€μˆ˜ λΆ„ν¬λŠ” 이산확λ₯ λΆ„포인 κΈ°ν•˜ λΆ„ν¬μ—μ„œ μ‹œν–‰ 간격을 κ·Ήν•œμœΌλ‘œ 쀄인 λ²„μ „μž…λ‹ˆλ‹€. 이것도 λ‚΄μš©μ΄ κΈΈμ–΄μ Έμ„œ 별도 포슀트둜 λΆ„λ¦¬ν•˜μ˜€μŠ΅λ‹ˆλ‹€ γ…Žγ…Ž

πŸ‘‰ Duality: Exponential Distribution and Geometric Distribution

맺음말

μ΄μ–΄μ§€λŠ” ν¬μŠ€νŠΈμ—μ„œλŠ” 연속 ν™•λ₯  λΆ„ν¬μ—μ„œ <μ •κ·œ 뢄포>λ§ŒνΌμ΄λ‚˜ μ€‘μš”ν•œ 뢄포인 <감마 뢄포; Gamma Distribution>에 λŒ€ν•΄ μ‚΄νŽ΄ λ³΄κ² μŠ΅λ‹ˆλ‹€ 🀩

πŸ‘‰ Gamma Distribution

References