Math Visualization List
Calculus
First Derivative Second Derivative Integration
Determinant
2x2 & 3x3 Determinants Cramer's Rule Permutations & Inversions Permutation Definition & Expansion Properties of Determinants: Row Operations Minors & Cofactors Two-Row Laplace Expansion Vandermonde Determinant
Matrix Operations
Matrix Addition, Subtraction & Scalar Multiplication Matrix Multiplication & Power Transpose, Determinant & Adjoint Inverse Matrix & Matrix Equations Elementary Transformations, Echelon & Canonical Form
Probability & Statistics
Conditional Probability, Independence & Total Probability Discrete Probability Distributions PMF & CDF 2D Random Vectors Conditional Probability Distributions Bayes' Theorem Numerical Characteristics of Random Variables Higher Moments & Covariance Matrix Law of Iterated Expectations Independence, Mean Independence & Uncorrelated Continuous Statistical Distributions
Vector Operations
Vector Addition, Subtraction & Scalar Multiplication Linear Combination, Span & Basis Matrices as Transformations Composition of Transformations 3D Linear Transformations Determinant Inverse, Column Space & Rank Cross-Dimensional Transformations Vector Angle Cross Product Change of Basis Eigenvectors & Eigenvalues Function Vector Spaces Geometric Interpretation of Cramer's Rule Gravity Simulation

Law of Iterated Expectations (LIE)

"The Expectation of the Conditional Expectation is the Expectation"

1. Mathematical Expression

$$E[X] = E[E[X|Y]]$$

Left $E[X]$: The unconditional expectation of random variable $X$ (big picture view).

Inner $E[X|Y]$: This is a random variable with respect to $Y$. When $Y$ takes different values, we observe the conditional average of $X$ (hierarchical view).

Outer $E[\cdot]$: Take the weighted average of these hierarchical means.

Discrete Form Expansion:

$$E[X] = \sum_{y} E[X|Y=y] \cdot P(Y=y)$$

This is the core formula demonstrated in the visualization below.

2. Intuitive Understanding: Hierarchical Thinking

Imagine you want to calculate the average height of all students in a school ($E[X]$):

  1. First, group students by grade (condition $Y$).
  2. Calculate the average height for each grade (conditional expectation $E[X|Y=y]$).
  3. Take the weighted average of these grade averages, weighted by grade size.
  4. The result equals the overall school average height.

Applications:

Econometrics Machine Learning Decision Theory Financial Engineering

3. Calculation Steps Breakdown

Step 1: Calculate Conditional Expectations

For each subgroup Y, calculate the mean of X:
E[X|A] = 40.0 E[X|B] = 70.0 E[X|C] = 55.0

Step 2: Get Weights

Probability weights for each subgroup:
P(A) = 30% P(B) = 50% P(C) = 20%

Step 3: Weighted Sum

40.0×0.30 + 70.0×0.50 + 55.0×0.20

Step 4: Get Unconditional Expectation

E[X] = 59.0

Multi-layer Population Expectation Decomposition

气泡大小代表权重 $P(Y)$, 内部高度代表 $E[X|Y]$

最终全期望 $E[X]$

0.00

Law of Iterated Expectations holds

Group A: E[X|Y=A]

0.0

30%
Group B: E[X|Y=B]

0.0

50%
Group C: E[X|Y=C]

0.0

20%

Verify Formula:

E[X] = Σ E[X|Y=y]·P(Y=y)

Key Insights

Significance of the Law

The Law of Iterated Expectations shows that we can calculate overall expectation by first computing conditional expectations at each level, then integrating them through weighted averaging. This provides a methodology for decomposing complex problems.

Visual Interpretation

In the visualization, each bubble's size represents the subgroup weight P(Y), and the column height represents the conditional expectation E[X|Y]. The baseline for E[X] shows where the weighted average falls.