Parameters and Moments of the Log-Normal
What you'll learn: How the log-normal distribution's parameters relate to the underlying normal distribution, and how to calculate its mean and variance.
Understanding the Parameters
The log-normal distribution has a clever twist: if a variable X is log-normally distributed, then ln(X) follows a normal distribution with parameters μ (mean) and σ (standard deviation).
Think of it this way: μ and σ aren't the mean and standard deviation of X itself—they describe the normal distribution of ln(X). This is like saying "the log of your data behaves normally with these specific values."
Key Parameter Facts
- μ: The mean of ln(X), not X
- σ: The standard deviation of ln(X), not X
- Both μ and σ can be any real numbers (μ can be negative, σ must be positive)
Calculating Mean and Variance of X
Here's where it gets interesting. The actual mean and variance of the log-normal variable X are:
Mean of X: E[X] = e^(μ + σ²/2)
Variance of X: Var(X) = e^(2μ + σ²) × (e^(σ²) - 1)
Notice how both involve exponential functions—that's because we're transforming back from the log scale. The mean depends on both μ and σ, and increasing σ actually increases the mean (the distribution becomes more right-skewed).
Why This Matters
In real applications like modeling income or stock prices, you often observe the actual values (X) but need to work with the underlying parameters (μ, σ) that describe the log-transformed data. Understanding this relationship lets you move between scales correctly.
Key Takeaway: The log-normal distribution is parameterized by μ and σ from the underlying normal distribution of ln(X), but its actual mean and variance involve exponential transformations of these parameters.