STA260 Lecture 14

Completed Notes Status

Central objective: Derive Beta distributions using Jacobian transformations and evaluate estimator performance through Mean Squared Error (MSE) and the variance-bias tradeoff.
Key concepts:
- Beta Distribution: Explored through variable transformation, showing how mapping $Y = X / θ$ and $Y = X^{1 / α}$ (when $X \sim U (0, 1)$ ) yields Beta distributions.
- Mean Squared Error: A universal metric for estimator goodness of fit, balancing the variance of an estimator with its squared bias.
- Variance-Bias Tradeoff: The principle that an estimator's overall error is composed of $Var (\hat{θ})$ and $b (\hat{θ})^{2}$ , and that minimizing MSE often requires trading off one against the other.
Connections:
- Variable transformations map basic probabilities into complex families (like Beta), while MSE provides a unified framework to mathematically evaluate the accuracy of the estimators derived from these underlying distributions.

#tk: practice this derivation from scratch
- Answer: Follow the Jacobian transformation mapping $X$ to $Y = X / θ$ , substituting into $f_{Y} (y) = f_{X} (g^{- 1} (y)) | J |$ until recognizing the $Beta (n, 1)$ kernel.
#tk: make flashcard of this shortcut.
- Answer: A flashcard has been created for the shortcut $X \sim U (0, 1), Y = X^{1 / α} ⟹ Y \sim Beta (α, 1)$ .
#tk: Remember this trick and context on how to use this.
- Answer: The trick is adding and subtracting the true parameter ( $μ$ ) inside the variance operator. This algebraically forces the presence of a standardized $Z$ score ( $Z = \frac{\bar{X} - μ}{σ / \sqrt{n}}$ ), letting you leverage the property $Z^{2} \sim χ_{(1)}^{2}$ (where variance is 2) instead of manually computing 4th moments.
#tk: an old final question
- Answer: The extensive algebraic derivation of $MSE [{\bar{X}}^{2}]$ utilizing the "add zero" trick was noted as a past final exam question.

Remember/Understand:
- What is the relationship between Mean Squared Error, variance, and bias?
- If $Z$ follows a standard normal distribution, what distribution does $Z^{2}$ follow?
- State the shortcut transformation mapping a Uniform distribution to a Beta distribution.
Apply/Analyze:
- Prove that while $\bar{X}$ is an unbiased estimator of $μ$ , ${\bar{X}}^{2}$ is a biased estimator of $μ^{2}$ .
- Derive the PDF of $Y$ given $X \sim U (0, 1)$ and $Y = X^{1 / α}$ .
Evaluate/Create:
- Evaluate why adding zero ( $μ - μ$ ) is necessary when deriving the variance of the squared sample mean ${\bar{X}}^{2}$ . What computational roadblocks does this bypass?

Mean Squared Error:
- Why it's challenging: Expanding $MSE [{\bar{X}}^{2}]$ requires manipulating variances of squared variables, handling 4th moments, and algebraic endurance.
- Study strategy: Practice the "add zero" trick explicitly, mapping each step strictly to the properties of standard normal variables and Chi-squared distributions without skipping lines.
Transformation of Variables:
- Why it's challenging: Maintaining track of the absolute value of the Jacobian $| J |$ and accurately grouping terms to recognize a target PDF kernel (like the Beta distribution) can easily result in arithmetic errors.
- Study strategy: Re-derive the Beta PDF starting from the order statistic $X_{(n)}$ without referencing the raw notes.

Immediate review actions:
- Review all [COMPLETED] insertions and verify with course materials
- Address all #tk items (if any)
- Re-read Lecture Summary and confirm key links
Practice and application:
- Answer Remember/Understand questions
- Answer Apply/Analyze questions
- Attempt Evaluate/Create questions
Deep dive study:
- Focus on Challenging Concepts (above)
- Extend atomic notes only where the new lecture adds content
Verification and integration:
- Cross-reference notes with textbook/slides
- Identify remaining gaps
- Link this lecture to prior/future lecture notes (lecture notes only; no MOCs)