Random Variables

Random Variables Study Guide


1. What is a Random Variable?

  • A random variable is a variable that takes on different values based on some probability distribution.

  • The set of all possible values is called the domain of the random variable.

  • If a random variable takes on a particular value, that event has an associated probability.

  • The distribution of a random variable is the collection of probabilities over all possible values.

Example: Rolling a Fair Die

  • Let D be a random variable representing the outcome of rolling a fair six-sided die.

  • The domain is {1,2,3,4,5,6}.

  • Each outcome has a probability of 1/6.


2. Functions of Random Variables

  • If we define a function f(V) over a random variable V, then f(V) itself is a random variable.

  • Example: If V is the outcome of a die roll and we define f(V) = 2V, then f(V) has the domain {2,4,6,8,10,12} with probabilities inherited from V.


3. Independent and Conditionally Independent Random Variables

Independence

  • Two random variables U and V are independent if:

    P(U = x, V = y) = P(U = x) * P(V = y) for all x, y

  • Example: Rolling two dice where D1 and D2 are the outcomes of the first and second roll, respectively.

Conditional Independence

  • Two random variables U and V are conditionally independent given Z if:

    P(U = x | Z = z, V = y) = P(U = x | Z = z) for all x, y, z

  • Example: Suppose a student's exam performance depends on study time S and sleep T, but given study time, sleep no longer affects the score. Then T and P (performance) are conditionally independent given S.


4. Stochastic Models

A stochastic model represents uncertainty in a system using probability.

Three Steps in Building a Stochastic Model:

  1. Choosing random variables – Identify which factors have randomness.

  2. Specifying independence/conditional independence assumptions – Define how variables interact.

  3. Determining probabilities – Assign values based on:

    • Empirical data (statistics from experiments)

    • Domain knowledge (e.g., a fair die has equal probabilities)

    • Guesstimates (informed assumptions)

    • Principle of Insufficient Reason (if no info, assume equal probabilities)


5. Joint Distributions

  • If two random variables U and V are considered together, their probabilities are represented by a joint distribution.

  • The joint probability distribution assigns probabilities to pairs of outcomes.

    P(U = x, V = y)

  • Example: Rolling a die (D) and flipping a coin (F), the outcomes are:

    (1H, 1T, 2H, 2T, ..., 6H, 6T)


6. Marginal Probability

  • Marginalizing over a random variable means summing over its possible values to get probabilities for another variable.

  • Formula:

    P(U = x) = sum over all y of P(U = x, V = y)

  • Example: Given a joint probability table for two variables, summing across rows or columns gives the marginal probability of each variable.


7. Expected Value

  • The expected value E[V] of a random variable V is:

    E[V] = sum over all x in domain(V) of x * P(V = x)

  • Example: Rolling a fair die

    E[D] = (1 1/6) + (2 1/6) + ... + (6 * 1/6) = 3.5


8. Expected Value Theorem

  • For any two random variables U and V:

    E[U + V] = E[U] + E[V]

  • Key property: Holds even if U and V are dependent.

Example: Counting Red Balls in an Urn

  • An urn has R red balls and B blue balls. We draw K balls at random.

  • Define N as the number of red balls drawn.

  • Using the expected value theorem:

    E[N] = K * (R / (R + B))


9. Practice Questions

  1. Multiple Choice

    • What is the expected value of rolling a fair die?

      • a) 3

      • b) 3.5

      • c) 4

      • d) 2.5

    • If X and Y are independent, which of the following is true?

      • a) P(X | Y) = P(X)

      • b) P(X, Y) = P(X) + P(Y)

      • c) P(X) = P(Y)

      • d) P(X, Y) = 1

  2. Short Answer

    • Define marginal probability and provide an example.

    • Show how to compute the expected value of rolling a biased die.

  3. Case Study

    • A factory machine produces parts with a 5% defect rate. If we sample 10 parts, what is the expected number of defective parts?