Functions of Random Variables: Continuous Case
You'll learn how to find the probability density function of a transformed continuous random variable using the CDF method.
The Core Idea
When you have a continuous random variable X with a known PDF, and you apply a function Y = g(X), you often need to find the PDF of this new random variable Y. The CDF method is a systematic approach:
- Start with the CDF of Y: F_Y(y) = P(Y ≤ y)
- Rewrite in terms of X: F_Y(y) = P(g(X) ≤ y)
- Solve the inequality for X
- Use the known CDF or PDF of X to evaluate
- Differentiate to get the PDF: f_Y(y) = dF_Y(y)/dy
Why This Works for Monotonic Functions
For monotonic transformations (strictly increasing or decreasing), the inequality g(X) ≤ y has a clear solution.
- If g is increasing: g(X) ≤ y becomes X ≤ g⁻¹(y)
- If g is decreasing: g(X) ≤ y becomes X ≥ g⁻¹(y)
Think of it like a one-way street: monotonic functions preserve or reverse order, so there's no ambiguity about which values of X satisfy the inequality.
Example Pattern
Suppose Y = 2X + 3 where X has PDF f_X(x). Then:
- F_Y(y) = P(2X + 3 ≤ y) = P(X ≤ (y-3)/2) = F_X((y-3)/2)
- Differentiate: f_Y(y) = f_X((y-3)/2) · (1/2)
Notice the Jacobian factor (1/2) appears from the chain rule—this compensates for the stretching or compression of the probability density.
Key Takeaway: The CDF method transforms Y = g(X) by expressing P(Y ≤ y) in terms of X's distribution, then differentiating; for monotonic g, this involves the inverse function and a Jacobian adjustment factor |dg⁻¹/dy|.