Course contentsShow
← Mathematics
Lesson 2929 of 3,12158. Joint and Limit BehaviorFree lesson

Marginal Distributions from Joint Distributions

Derive marginal probability distributions by summing or integrating the joint distribution over the other variable(s).

Marginal Distributions from Joint Distributions

What you'll learn: How to extract the probability distribution of a single variable from a joint distribution by "marginalizing out" the other variable(s).

The Core Idea

When you have a joint probability distribution describing two or more random variables together, you often need to find the distribution of just one of those variables on its own. This individual distribution is called a marginal distribution.

Think of it like viewing a 3D object from different angles—if you shine a light from above and look at the shadow on the ground, you're "projecting" the full 3D shape onto a 2D plane. Similarly, marginalizing collapses a multi-variable distribution down to focus on just one variable.

How It Works

For Discrete Random Variables

If X and Y have joint PMF p(x, y), the marginal PMF of X is:

p_X(x) = Σ_y p(x, y)

You sum over all possible values of Y. This accounts for "all the ways X can equal x, regardless of what Y does."

For Continuous Random Variables

If X and Y have joint PDF f(x, y), the marginal PDF of X is:

f_X(x) = ∫_{-∞}^{∞} f(x, y) dy

You integrate out Y over its entire range. This accumulates all the probability density along the "Y-axis" at each fixed value of x.

Example Intuition

Imagine a table showing joint probabilities of weather (sunny/rainy) and traffic (light/heavy). To find the marginal probability of just "sunny weather," add up all scenarios where it's sunny: sunny-light + sunny-heavy. That gives you the probability of sunny weather overall, ignoring traffic conditions.

Key Takeaway: Marginal distributions are obtained by summing (discrete) or integrating (continuous) the joint distribution over all values of the unwanted variable(s), leaving only the variable of interest.