Variance Estimation
โํ๋ฅ ๊ณผ ํต๊ณ(MATH230)โ ์์ ์์ ๋ฐฐ์ด ๊ฒ๊ณผ ๊ณต๋ถํ ๊ฒ์ ์ ๋ฆฌํ ํฌ์คํธ์ ๋๋ค. ์ ์ฒด ํฌ์คํธ๋ Probability and Statistics์์ ํ์ธํ์ค ์ ์์ต๋๋ค ๐ฒ
Interval Estimation ํฌ์คํธ์์ ๋ค๋ฃฌ <Interval Estimation>์ ํน์ ์ํฉ์ ์ด๋ป๊ฒ ์ ์ฉํ ์ ์๋์ง๋ฅผ ๋ค๋ฃจ๋ ํฌ์คํธ์ ๋๋ค.
Single Sample Estimation: Variance Estimation
Let $X_1, \dots, X_n$ be a random sample from $N(\mu, \sigma^2)$.
Q. Find $100(1-\alpha)\%$ confidence interval for $\sigma^2$.
1. choose a point estimator for $\sigma^2$.
\[S^2 = \frac{1}{n-1} \cdot \sum^n_{i=1} (X_i - \bar{X})^2\]and also
\[\frac{(n-1) S^2}{\sigma^2} \; \sim \; \chi^2 (n-1)\]2. Find confidence interval by using $\dfrac{(n-1) S^2}{\sigma^2} \; \sim \; \chi^2 (n-1)$.
\[1 - \alpha = P \left( \chi^2_{1-\alpha/2} (n-1) \; \le \; \frac{(n-1)s^2}{\sigma^2}\; \le \; \chi^2_{\alpha/2} (n-1)\right)\]Therefore, the confidence interval for $\sigma^2$ is
\[\left( \frac{(n-1)\cdot s^2}{\chi^2_{\alpha/2}}, \frac{(n-1)\cdot s^2}{\chi^2_{1 - \alpha/2}} \right)\]where $\chi^2_{\alpha/2}$ and \(\chi^2_{1-\alpha/2}\) are $\chi^2$-values with $n-1$ dof.
๐ฅ NOTE: $X_1, \dots, X_n$ should follow $N(\mu, \sigma^2)$.
๐ฅ NOTE: $\chi^2$ distribution is NOT symmetric!
Two Samples Estimation: The ratio of two variances
Let $X_1, \dots, X_{n_1}$ be random samples from $N(\mu_1, \sigma_1^2)$.
Let $Y_1, \dots, Y_{n_2}$ be random samples from $N(\mu_2, \sigma_2^2)$.
Supp. $X_i$s and $Y_j$s are independent.
Q. Find $100(1-\alpha)\%$ confidence interval for $\sigma_1^2 / \sigma_2^2$.
A. We can use $s_1^2 / s_2^2$ instead!!
\[\frac{s_1^2 / \sigma_1^2}{s_2^2 / \sigma_2^2} \; \sim \; F(n_1-1, n_2-1)\]๋ฐ๋ผ์,
\[1 - \alpha = P \left( f_{1-\alpha/2} \, (n_1 - 1, n_2 - 1) \; \le \; \frac{s_1^2 / \sigma_1^2}{s_2^2 / \sigma_2^2} \; \le \; f_{\alpha/2} \, (n_1 - 1, n_2 - 1) \right)\]Note that $f_{1-\alpha/2} \, (n_1 - 1, n_2 - 1) = \dfrac{1}{f_{\alpha/2} \, (n_2 - 1, n_1 - 1)}$.
Therefore, the confidence interval for $\sigma_1^2 / \sigma_2^2$ is
\[\left( \frac{s_1^2}{s_2^2} \cdot \frac{1}{f_{\alpha/2} \, (n_1 - 1, n_2 - 1)}, \; \frac{s_1^2}{s_2^2} \cdot f_{\alpha/2} \, (n_2 - 1, n_1 - 1) \right)\]์ง๊ธ๊น์ง โ์ถ์ (Statistical Estimation)โ ๊ณผ์ ์ ๋ํด ์ดํด๋ณด์๋ค. $\bar{x}$์ $s^2$์ ๊ฐ์ด Point Estimator๋ฅผ ๊ตฌํ๋ ๊ฒฝ์ฐ๋ ์์๊ณ , $\bar{x} \pm z_{\alpha/2} \cdot s / \sqrt{n}$ ๊ณผ ๊ฐ์ด Interval Estimator๋ฅผ ๊ตฌํ๋ ๊ฒฝ์ฐ๋ ์์๋ค. ๋, Single Sample์์ Estimator๋ฅผ ๊ตฌํ๋ ๊ฒ๋ ์์๊ณ , Two Samples์์ Estimator๋ฅผ ๊ตฌํ๋ ๊ฒ๋ ์์๋ค.
๊ทธ๋ฌ๋ ์ง๊ธ๊น์ง ์ดํด๋ณธ ๋ฐฉ์ ์ธ์๋ ๋๋ค๋ฅธ Estimation ๋ฐฉ๋ฒ์ด ์๋ค ๐ฒ <MLE; Maximum Likelihood Estimation>๋ Sample Distribution์ ์ฌํํ ํ๋ฅ ์ด ๊ฐ์ฅ ๋์ Parameter $\theta$๋ฅผ ์ฐพ๋ ๋ฐฉ์์ผ๋ก Estimator๋ฅผ ์ฐพ๋๋ค. <MLE>๋ ๋น์ฐํ Point Estimator๋ฅผ ์ ์ํ๋ฉฐ, ๊ทธ ๊ณผ์ ์์ $\theta$์ ๋ํด ํธ๋ฏธ๋ถ์ ์ํํ๋ค.