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

3 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

Sample Mean Test

ํ‰๊ท (Mean)์— ๋Œ€ํ•œ ๊ฒ€์ •์€ ์ถ”์ •์—์„œ์™€ ๋น„์Šทํ•˜๊ฒŒ, $\sigma^2$์„ ์•„๋Š”์ง€ ์—ฌ๋ถ€์— ๋”ฐ๋ผ ๋‹ค๋ฅด๊ฒŒ ์ ‘๊ทผํ•œ๋‹ค.


$\sigma^2$ is known

1. ์ƒํ™ฉ ์ธ์‹

  • $H_0: \mu=190$
  • $H_1: \mu > 190$

$n=25$ and $\bar{x}=194$, and $\sigma^2 = 100$

2. Find a <Test Statistic>, and construct critical region

  • Test Statistic: $\bar{x}$
  • critical region: $\{ \bar{X} > C\}$

3. $\alpha$๊ฐ€ ์ฃผ์–ด์ง€์ง€ ์•Š์•˜์œผ๋‹ˆ, p-value๋ฅผ ๊ตฌํ•˜์ž!

\[\begin{aligned} \alpha &= P(\bar{X} \ge 194 \mid \mu = 190) \\ &= P \left(\frac{\bar{X} - 190}{\sigma / \sqrt{n}} \ge \frac{194 - 190}{\sigma / \sqrt{n}} \right) \\ &= P(Z \ge 2) = 0.023 \end{aligned}\]

4. ๊ฒฐ์ •

  • If $\alpha > 0.023$, reject $H_0$
  • If $\alpha < 0.023$, fail to reject $H_0$

$H_1: \mu < \mu_0$์ธ ๊ฒƒ๋„, $H_1: \mu \ne \mu_0$ ๊ฒฝ์šฐ๋„ ๋น„์Šทํ•˜๊ฒŒ, ์‹์„ ์ž˜ ์„ธ์›Œ์„œ ์ง„ํ–‰ํ•˜๋ฉด ๋œ๋‹ค!

์ •๋ฆฌํ•˜๋ฉด ์•„๋ž˜์™€ ๊ฐ™๋‹ค.

Quick Remark.

์‚ฌ์‹ค, โ€œrejection regionโ€๊ณผ โ€œconfidence intervalโ€์˜ ์„œ๋กœ Complementํ•˜๋‹ค!!

๋งŒ์•ฝ, ์šฐ๋ฆฌ๊ฐ€ ์–ป์€ sample mean $\bar{x}$์ด $H_0$์—์„œ ๊ฐ€์ •ํ•œ $\mu$์˜ confidence interval์— ํฌํ•จ๋œ๋‹ค๋ฉด,

\[\bar{x} \in (\mu \pm z_{\alpha/2} \cdot \sigma/\sqrt{n}) \quad \text{or} \quad \bar{x} < \mu + z_{\alpha} \cdot \sigma/\sqrt{n} \quad \text{or} \quad \bar{x} > \mu - z_{\alpha} \cdot \sigma/\sqrt{n}\]

์šฐ๋ฆฌ๋Š” $H_0$์„ ๊ธฐ๊ฐํ•  ์ด์œ ๊ฐ€ ์—†๋‹ค. ํ•˜์ง€๋งŒ, ๋งŒ์•ฝ $\bar{x}$๊ฐ€ confidence interval์„ ๋ฒ—์–ด๋‚œ๋‹ค๋ฉด, ์šฐ๋ฆฌ๋Š” ์šฐ๋ฆฌ๊ฐ€ ์„ค์ •ํ•œ $\mu$ ๊ฐ’์„ ์˜์‹ฌํ•˜๊ณ , ๋˜ ๊ธฐ๊ฐํ•ด์•ผ ํ•œ๋‹ค.

์ด๊ฒƒ์€ ์ฆ๋ช… ๋ฐฉ์‹ ์ค‘ ํ•˜๋‚˜์ธ โ€œ๊ท€๋ฅ˜๋ฒ•โ€๊ณผ ์œ ์‚ฌํ•œ๋ฐ, โ€œํ†ต๊ฒŒ์ • ๊ฒ€์ •(Testing)โ€์€ โ€œํ™•๋ฅ โ€์„ ์‚ฌ์šฉํ•ด ์ฒ˜์Œ์˜ ๊ฐ€์ • $H_0$๋ฅผ ๊ธฐ๊ฐํ•œ๋‹ค๊ณ  ๋ณผ ์ˆ˜ ์žˆ๋‹ค!


$\sigma^2$ is unknown

๋งŒ์•ฝ, $\sigma^2$๋ฅผ ๋ชจ๋ฅธ๋‹ค๋ฉด, ์ถ”์ •์—์„œ ํ–ˆ๋˜ ๊ฒƒ์ฒ˜๋Ÿผ <t-test>๋ฅผ ์ง„ํ–‰ํ•˜๋ฉด ๋œ๋‹ค. ๋„ˆ๋ฌด ์‰ฌ์šฐ๋‹ˆ ์„ค๋ช…์€ ์ƒ-๋žต ํ•˜๊ฒ ๋‹ค.

์ •๋ฆฌํ•˜๋ฉด ์•„๋ž˜์™€ ๊ฐ™๋‹ค.

๐Ÿ’ฅ ์ฃผ์˜!! ์ƒ˜ํ”Œ์€ ๋ฐ˜.๋“œ.์‹œ. Normal Distribution์—์„œ ์ถ”์ถœ๋˜์–ด์•ผ ํ•œ๋‹ค!!


Two Samples Mean Test

์ด๊ฒƒ๋„ ์‚ฌ์‹ค ๋ณ„๊ฑฐ ์—†๋‹ค. ๊ทธ๋ƒฅ ์ถ”์ •ํ•ด์„œ ํ–ˆ๋˜ ๊ฒƒ๊ณผ ์•ž์—์„œ ํ–ˆ๋˜ ๊ฒƒ์„ ์ž˜ ๋…น์—ฌ์„œ ๊ฒ€์ •์„ ์ˆ˜ํ–‰ํ•˜๋ฉด ๋œ๋‹ค.

์ •๋ฆฌํ•˜๋ฉด ์•„๋ž˜์™€ ๊ฐ™๋‹ค.


๋งˆ์ฐฌ๊ฐ€์ง€ ๋ฐฉ๋ฒ•์œผ๋กœ <Test for Paired Observations>์—์„œ๋„ ๊ทธ.๋Œ€.๋กœ. ์ž˜ ์ˆ˜ํ–‰ํ•˜๋ฉด ๋œ๋‹ค ๐Ÿ˜



๋‹ค์Œ ํฌ์ŠคํŠธ์—์„œ๋Š” <๊ฒ€์ •๋ ฅ; power of test> $\beta$๋ฅผ ๋„์ž…ํ•ด ํ‰๊ท (Mean)์— ๋Œ€ํ•ด ๊ฒ€์ •(Testing)์„ ์ˆ˜ํ–‰ํ•  ๋•Œ ํ•„์š”ํ•œ Sample Size $n$์„ ๊ฒฐ์ •ํ•˜๋Š” ๋ฐฉ๋ฒ•์— ๋Œ€ํ•ด ์‚ดํŽด๋ณธ๋‹ค.

๐Ÿ‘‰ Choice of Sample Size for Testing Mean


์ด๋ฒˆ ํฌ์ŠคํŠธ์—์„œ๋Š” ํ‰๊ท (Mean)์— ๋Œ€ํ•œ ๊ฒ€์ • ๋ฐฉ๋ฒ•์— ๋Œ€ํ•ด ์‚ดํŽด๋ดค๋‹ค. ๊ทธ๋ ‡๊ฒŒ ์–ด๋ ต์ง€ ์•Š์•˜๊ณ , ์ถ”์ •(Estimation)์—์„œ ํ•˜๋˜ ๊ฑธ, ์ ˆ์ฐจ์— ๋งž๊ฒŒ ์ˆ˜ํ–‰ํ•ด ํ•ด์„ํ•ด์ฃผ๋ฉด ๋˜๋Š” ๊ฑฐ์˜€๋‹ค. ์ด์–ด์ง€๋Š” ํฌ์ŠคํŠธ์—์„œ๋Š” ๋น„์œจ(proportion)๊ณผ ๋ถ„์‚ฐ(variance)์— ๋Œ€ํ•œ ๊ฒ€์ •์„ ์‚ดํŽด๋ณธ๋‹ค!

๐Ÿ‘‰ Test on Proportion and Variance