91Ó°ÊÓ

Expected value is fundamentally the average you expect from a random variable after many trials. It's a useful property to predict outcomes in probability and statistics. Notably, expected value is linear, which means it follows these rules:

• Additivity: The expected value of a sum of variables is the sum of their expected values. Formally, for random variables \(X\) and \(Y\), we have \(E[X + Y] = E[X] + E[Y]\).
• Scalar Multiplication: Multiplying a random variable by a constant scales its expected value by the same constant. If \(a\) is a constant, then \(E[aX] = aE[X]\).

This linearity helps in simplifying expressions involving expectations. When working with transformed random variables, such as \(aX + b\), the expected value becomes \(E[aX + b] = aE[X] + b\). You can see how straightforward it is to compute new expectations using these rules.

Variance measures how much a set of numbers, like the outcomes of a random variable, are spread out. It gives an idea of the variability or consistency of the data. The formula is:\[\operatorname{Var}(X) = E[(X - E[X])^2]\]Variance can be adjusted when random variables are transformed:

• For \(aX + b\), \(\operatorname{Var}(aX + b) = a^2 \operatorname{Var}(X)\).

The standard deviation \(\sigma\), being the square root of the variance, measures spread in the original units of the data:

• Standard deviation for \(aX + b\) is \(|a|\sigma_X\).

Understanding these transformations is crucial when analyzing how data changes under different conditions or manipulations.

The correlation coefficient, denoted by \(\operatorname{Corr}(X, Y)\), measures the strength and direction of a linear relationship between two random variables \(X\) and \(Y\). It is computed as:\[\operatorname{Corr}(X, Y) = \frac{\operatorname{Cov}(X, Y)}{\sigma_X \sigma_Y}\]This value ranges between -1 and 1:

• A value of 1 means a perfect positive linear relationship.
• -1 indicates a perfect negative linear relationship.
• 0 suggests no linear relationship.

For transformed variables like \(aX + b\) and \(cY + d\), the correlation remains \(\operatorname{Corr}(X, Y)\) when \(a\) and \(c\) are consistent in sign. If \(a\) and \(c\) have opposite signs, the correlation becomes the negative of \(\operatorname{Corr}(X, Y)\). This shows how scaling and translating variables affect their linear relationship.

Transforming random variables is common in statistics, whether for scaling, translating, or more complex operations. The linear transformation of a random variable \(X\) can be expressed as \(aX + b\). Here's why we transform:

• Simplifying complex data structures.
• Aligning variables to a common scale or unit.
• Improving data interpretability.

When dealing with transformations, it's important to know how properties like expected value, variance, and correlation change. As noted earlier, variance becomes \(a^2\operatorname{Var}(X)\) and standard deviation changes to \(|a|\sigma_X\). Correlation, notably, can flip its sign depending on the relationship between the constants used (positive or negative). Transformations thus help us manipulate data for better analysis without altering fundamental statistical relationships.