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

8 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

Probability


Definition. Probability of an event $A$; $P(A)$

Let $S$ be a sample space. The probability of an event $A$ is the sum of the probabilities of all sample points in $A$.

And there are following properties:

  1. $0 \le P(A) \le 1$ for every event $A$
  2. $P(S) = 1$
  3. If $A \cap B = \emptyset$, then $P(A \cup B) = P(A) + P(B)$
  4. If $A_1$, $A_2$, โ€ฆ is a sequence of mutually exclusive events, then $P(A_1 \cap A_2 \cap \cdots) = P(A_1) + P(A_2) + \cdots$


Term. Equally likely outcomes

<Equally likely outcomes> mean that each element in the sample space occurs with equal chance.

Under this circumstance, if $S = \{ x_1, \dots, x_N \}$ so that $\left| S \right| = N$, then we define $P(A) := \dfrac{\left| A \right|}{N}$.


Additive Rule


Theorem.

For any two events $A$ and $B$,

$P(A \cup B) = P(A) + P(B) - P(A \cap B)$


proof.

Exercise

// ๋ณธ์ธ์€ โ€œProbabilityโ€๋ฅผ ๊ณ„์‚ฐํ•  ๋•Œ, ๊ทธ event์— ์†ํ•œ sample point์˜ โ€œProbabilityโ€๋ฅผ ํ•ฉํ•œ๋‹ค๊ณ  ์ƒ๊ฐํ•จ. ๊ทธ๋Ÿฐ๋ฐ ๋งŒ์•ฝ $P(A) + P(B)$๋งŒ ํ•˜๊ฒŒ ๋˜๋ฉด, $A \cap B$์— ์†ํ•˜๋Š” sample point์˜ ํ™•๋ฅ ์„ ์ค‘๋ณตํ•ด์„œ ๋”ํ•˜๋Š” ๊ผด์ด ๋˜๊ธฐ ๋•Œ๋ฌธ์— ์ด๊ฒƒ์„ ์ œ์™ธํ•ด์ค˜์•ผ ํ•œ๋‹ค๊ณ  ์ƒ๊ฐํ•จ.


Topic. Matching Problem

์ˆ˜ํ•™ ์‹œํ—˜์—์„œ 3๋ช…์˜ ํ•™์ƒ๋“ค์ด ์ž์‹ ๋“ค์ด ์นœ ์‹œํ—˜์ง€๋ฅผ ์ฑ„์ ํ•œ๋‹ค๊ณ  ํ•œ๋‹ค. ์„ ์ƒ์€ ํ•™์ƒ๋“ค์ด ์ž๊ธฐ ์ž์‹ ์˜ ์‹œํ—˜์ง€๋ฅผ ์ฑ„์ ํ•˜์ง€๋Š” ์•Š๋„๋ก ํ•˜๊ณ  ์‹ถ๋‹ค. ๊ทธ ํ™•๋ฅ ์€ ์–ด๋–ป๊ฒŒ ๋˜๋Š”๊ฐ€?

// ๋ณธ์ธ์€ ์ฒ˜์Œ ๋ฌธ์ œ๋ฅผ ํ’€์—ˆ์„ ๋•Œ, ํ‹€๋ ธ์—ˆ๋‹ค ใ… ใ… 


Conditional Probability


Defitnition. Conditional Probability

The conditional probability of $B$, given $A$, denoted by $P(B \mid A)$, is defined by

\[P(B \mid A) = \frac{P(B \cap A)}{P(A)}, \quad \mbox{provided} \; P(A) > 0\]

โ€œThe notion of conditional probability provides the capability of reevaluating the idea of probability of an event in light of additional informationโ€

The probability $P(A \mid B)$ is an updating of $P(A)$ based on the knowledge that event $B$ has occurred.

Independent Events


Definition. Independent

Two events $A$ and $B$ are independent if and only if

\[P(B \mid A) = P(B) \quad \mbox{or} \quad P(A \mid B) = P(A)\]

assuming the existences of the conditional probabilities.

Otherwise, $A$ and $B$ are dependent.

If two events $A$ and $B$ are independent, then the occrurence of $B$ had no impact on the odds of occurrence of $A$.


Product Rule


Theorem.

If an experiment the events $A$ and $B$ can both occur, then

\[P(A \cap B) = P(A) P(B \mid A), \quad \mbox{providied} \; P(A) > 0\]

Codntional Probability์™€ ํ•จ๊ป˜ Product Rule์˜ ์˜๋ฏธ๋ฅผ ๊ณฑ์”น์–ด ๋ณด์ž.

$P(A \cap B)$๊ฐ€ $P(A)$์™€ $P(B \mid A)$์˜ ๊ณฑ์œผ๋กœ ํ‘œํ˜„๋œ๋‹ค๊ณ  ํ•œ๋‹ค. ์ฆ‰, โ€œ$A$๊ฐ€ ๋ฐœ์ƒํ•  ํ™•๋ฅ โ€ $P(A)$์— โ€œ$A$๊ฐ€ ๋ฐœ์ƒํ–ˆ์„ ๋•Œ, $B$๊ฐ€ ๋ฐœ์ƒํ•  ํ™•๋ฅ  $P(B \mid A)$โ€๋ฅผ ๊ณฑํ•ด์ฃผ๋Š” ๊ฑฐ๋‹ค.

๋‹ค์‹œ ๋งํ•˜๋ฉด, ๋‘ ์‚ฌ๊ฑด $A$, $B$์— ๋Œ€ํ•ด, ๊ทธ ๋‘˜์ด ๋™์‹œ์— ๋ฐœ์ƒํ•˜๋Š” ์‚ฌ๊ฑด $A \cap B$๋ฅผ $A$ ๋ฐœ์ƒ ํ›„ $B$๊ฐ€ ๋ฐœ์ƒํ•œ ์‚ฌ๊ฑด์œผ๋กœ ํ•ด์„ํ•˜๋Š” ์…ˆ์ด๋‹ค. ์ด๋•Œ, $A$๊ฐ€ ๋ฐœ์ƒํ–ˆ๋‹ค๋ฉด, ๊ทธ ์ •๋ณด๋ฅผ ์‚ฌ๊ฑด $B$ ๋ฐœ์ƒ์— ๋ฐ˜์˜ํ•ด์•ผ ํ•˜๊ธฐ ๋•Œ๋ฌธ์— $P(B \mid A)$๋ผ๋Š” conditional probability๋ฅผ ๋„์ž…ํ•œ ๊ฒƒ์ด๋‹ค.


Theorem.

Two events $A$ and $B$ are independent if and only if

\[P(A \cap B) = P(A) P(B)\]


Notation. $A \perp B$

When two events $A$ and $B$ are independent, we denote it as

\[A \perp B\]


Statements.

1. If $A \perp B$, then can $A \perp Bโ€™$?

2. If $A \perp B$, $B \perp C$, and $C \perp A$, then $A \perp (B \cap C)$?

3. If $A \perp B$ and $B \perp C$, then can $A \perp C$?

4. If $A \cap B = \emptyset$, then $A \perp B$?

5. If $A$ is independent of all, and also independent to $A$ itself. What can be $P(A)$?

์ •๋‹ต ๋ณด๊ธฐ

1. Yes. We know $P(A \cap B) + P(A \cap Bโ€™) = P(A)$, and $P(A \cap B) = P(A)P(B)$. ์ด ๋‘ ์‹์„ ์ž˜ ์ •๋ฆฌํ•˜๋ฉด, $P(A \cap Bโ€™) = P(A)P(Bโ€™)$๋ฅผ ์–ป์„ ์ˆ˜ ์žˆ๋‹ค!


2. No. ๋ฐ˜๋ก€๋ฅผ ์ฐพ์„ ์ˆ˜ ์žˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด ๋™์ „ ๋‘๊ฐœ๋ฅผ ๋˜์ ธ H/T๋ฅผ ๊ธฐ๋กํ•˜๋Š” Sample Space๋ฅผ ์ƒ๊ฐํ•ด๋ณด์ž. ๊ทธ๋ฆฌ๊ณ  Event $A$, $B$, $C$๋ฅผ ์•„๋ž˜์™€ ๊ฐ™์ด ์ •์˜ํ•˜์ž.

\[A = \{HT, TH\} \quad B =\{HT, HH\}, \quad C = \{HT, TT\}\]

ํ™•์ธ์„ ํ•ด๋ณด๋ฉด, $A$, $B$, $C$๋Š” pairwise independent ํ•˜๋‹ค๋Š” ๊ฑธ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.

ํ•˜์ง€๋งŒ, $A$์™€ $B \cap C$๊ฐ€ independentํ•œ์ง€ ํ™•์ธํ•ด๋ณด์ž.

\[P(A \cap (B \cap C)) = \frac{1}{4} \ne P(A)P(B \cap C)\]

์ฆ‰, $A$์™€ $B \cap C$๋Š” dependentํ•˜๋‹ค! source


3. No. ์œ„์˜ ์˜ˆ์‹œ์—์„œ ์•ฝ๊ฐ„๋งŒ ๋ณ€ํ˜•ํ•˜๋ฉด ์‰ฝ๊ฒŒ ๋ฐ˜๋ก€๋ฅผ ์ฐพ์„ ์ˆ˜ ์žˆ๋‹ค!!

\[A = \{HT, TH\} \quad B =\{HT, TT\}, \quad C = \{HH, TT\}\]

ํ™•์ธ์„ ํ•ด๋ณด๋ฉด, $A \perp B$, $B \perp C$์ธ ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.

ํ•˜์ง€๋งŒ, $A \cap C = \emptyset$์ด๊ธฐ ๋•Œ๋ฌธ์— $P(A \cap C) \ne P(A)P(C)$์ด๋‹ค!


4. No. ๋ฐ˜๋ก€๋Š” ๋„ˆ๋ฌด ๊ฐ„๋‹จํ•ด์„œ ์ƒ๋žต


5. $P(A) = 1$ or $P(A) = 0$. ๊ฐ„๋‹จํ•œ ๋Œ€์ˆ˜์‹์„ ํ’€๋ฉด ๋œ๋‹ค. โ€œindependent to $A$ itselfโ€๊ฐ€ ํžŒํŠธ์ธ๋ฐ, $P(A \cap A) = P(A)P(A)$์ด๋ฏ€๋กœ

\[P(A \cap A) = P(A) = P(A)P(A)\]

๋ฅผ ํ’€๋ฉด ๋œ๋‹ค. ํ™•๋ฅ ์˜ ์ •์˜์— ๋”ฐ๋ผ $0 \le P(A) \le 1$์ด๋ฏ€๋กœ ่งฃ๋Š” $P(A) = 1$ or $P(A) = 0$์ด ๋œ๋‹ค.


์ด์–ด์ง€๋Š” ๋‚ด์šฉ์€ ์ •๋ง์ •๋ง ์ค‘์š”ํ•˜๊ณ , ์œ ์šฉํ•œ <๋ฒ ์ด์ฆˆ ๊ทœ์น™ Bayesโ€™ Rule>์ด๋‹ค!!