Bivariate Distributions | 二元分布

📚 Bivariate Distributions | 二元分布

In statistics and probability, many real-world phenomena involve two random variables that may be related. A bivariate distribution describes the joint behaviour of two random variables, X and Y. Understanding such distributions is crucial for exploring dependence, correlation, and conditional relationships. In IB Mathematics, particularly within the Analysis and Approaches and Applications and Interpretation syllabi, bivariate distributions appear in topics such as joint probability mass functions, joint probability density functions, covariance, correlation, and the bivariate normal distribution. This article provides a comprehensive yet exam-focused guide to bivariate distributions, covering both discrete and continuous cases.

在统计学和概率论中，许多现实世界的现象都涉及两个可能相关的随机变量。二元分布描述了随机变量 X 和 Y 的联合行为。理解这类分布对于探究依赖性、相关性和条件关系至关重要。在 IB 数学（特别是 Analysis and Approaches 与 Applications and Interpretation 课程）中，二元分布出现在联合概率质量函数、联合概率密度函数、协方差、相关性以及二元正态分布等主题中。本文提供一份全面且紧扣考点的二元分布指南，涵盖离散和连续两种情形。

1. What is a Bivariate Distribution? | 什么是二元分布？

A bivariate distribution concerns two random variables, X and Y, defined on the same sample space. It assigns probabilities to pairs of outcomes (x, y). If both X and Y are discrete, the distribution is described by a joint probability mass function (PMF). If they are continuous, a joint probability density function (PDF) is used. Bivariate distributions allow us to study how the variables interact, whether they are independent, and how to predict one from the other.

二元分布涉及定义在同一个样本空间上的两个随机变量 X 和 Y。它为成对的结果 (x, y) 分配概率。如果 X 和 Y 都是离散的，该分布由联合概率质量函数（PMF）描述；如果它们都是连续的，则使用联合概率密度函数（PDF）。二元分布使我们能够研究变量如何相互作用，它们是否独立，以及如何用一个变量预测另一个变量。

---

2. Discrete Joint Probability Distributions | 离散联合分布

be presented in a two-way table, making it easy to read probabilities and compute marginal sums.

对于离散随机变量 X 和 Y，联合概率质量函数定义为 P(X = x, Y = y)。它必须满足两个条件：对所有 (x, y)，0 ≤ P(x, y) ≤ 1，并且所有可能配对的和 ΣΣ P(x, y) = 1。联合 PMF 可以用二维表格呈现，便于读取概率和计算边缘总和。

Example: e first coin shows Heads, 0 otherwise. Let Y be the total number of Heads. The joint PMF table is:

例子：考虑两枚公平硬币。令 X = 1 表示第一枚为正面，否则为 0。令 Y 表示正面总数。联合 PMF 表格如下：

X \\ Y	0	1	2
0	0.25	0.25	0
1	0	0.25	0.25

e sum of all cell probabilities is 1. Such tabular representation is an excellent way to visualise the joint behaviour and to verify the normalisation condition.

这里每个单元格表示 P(X = x, Y = y)。所有单元格概率之和为 1。这种表格表示是可视化联合行为并验证归一化条件的极好方法。

---

3. Continuous Joint Probability Distributions | 连续联合分布

, y) is used. Probabilities are found by integrating the joint PDF over a region: P((X, Y) ∈ A) = ∬ oint PDF must satisfy f(x, y) ≥ 0 and the total integral over the entire plane equals 1: ∫_-∞^∞ ∫_-∞^∞ f(x, y) dx dy = 1.

当 X 和 Y 为连续型时，使用联合概率密度函数 f(x, y)。概率通过对联合 PDF 在某个区域上积分求得：P((X, Y) ∈ A) = ∬_A f(x, y) dx dy。联合 PDF 必须满足 f(x, y) ≥ 0，并且整个平面上的积分为 1：∫_-∞^∞ ∫_-∞^∞ f(x, y) dx dy = 1。

For example, suppose f(x, y) = kxy for 0 < x < 2 and 0 < y < 1, and zero otherwise. Then ∫₀¹ ∫₀² udents must carefully define the integration region and set up double integrals correctly.

常见的考题是求常数 k，使给定函数成为有效的联合 PDF。例如，设 f(x, y) = kxy，0 < x < 2，0 < y < 1，其余为零。则 ∫₀¹ ∫₀² kxy dx dy = 1，解得 k = 1。学生必须小心定义积分区域并正确建立二重积分。

---

4. Marginal Distributions | 边缘分布

X is P(X = x) = Σ_y P(X = x, Y = y). For continuous variables, the marginal PDF of X is f_X(x) = ∫_-∞^∞ f(x, y) dy. Similarly for Y. Marginals allow us to study one variable while 'summing out' or 'integrating out' the other.

边缘分布从联合分布中恢复出 X 或 Y 的单独行为。对于离散变量，X 的边缘 PMF 为 P(X = x) = Σ_y P(X = x, Y = y)。对于连续变量，X 的边缘 PDF 为 f_X(x) = ∫_-∞^∞ f(x, y) dy。类似地可求得 Y 的边缘分布。边缘分布使我们能够在“求和掉”或“积分掉”另一个变量的情况下研究某一变量。

.5, as expected for a fair coin. Marginal distributions must also sum (or integrate) to 1.

在上述硬币例子中，X 的边缘分布：P(X=0) = 0.25+0.25+0 = 0.5，P(X=1) = 0+0.25+0.25 = 0.5，正如对公平硬币的预期。边缘分布也必须满足总和（或积分）为 1。

---

5. Conditional Distributions | 条件分布

ibution of one variable given a specific value of the other. In the discrete case, P(Y = y | , provided P(X = x) > 0. In the continuous case, the conditional PDF of Y given X = x is f_Y|X(y|x) = f(x, y) / f_X(x). These are powerful tools for prediction and for understanding dependence structures.

条件分布描述在给定一个变量特定值时另一个变量的概率分布。在离散情形，P(Y = y | X = x) = P(X = x, Y = y) / P(X = x)，前提是 P(X = x) > 0。在连续情形，给定 X = x 下 Y 的条件 PDF 为 f_Y|X(y|x) = f(x, y) / f_X(x)。这些是进行预测和理解依赖结构的有力工具。

Conditional expectations like E(Y | rating y times the conditional distribution. These form the basis of regression curves and are heavily tested in IB exam questions involving bivariate normal or simple discrete joint tables.

条件期望，如 E(Y | X = x)，通过对 y 乘以条件分布求和或积分来计算。它们构成了回归曲线的基础，并在涉及二元正态或简单离散联合表格的 IB 考题中频繁出现。

---

广告 / Sponsored

6. Independence of Random Variables | 随机变量的独立性

ns P(X = x, Y = y) = P(X = x) P(Y = y) for all x, y. For continuous variables, f(x, y) = f_X(x) f_Y(y) for all x, y. Independence implies that knowing the value of one variable gives no information about the other. Checking independence in a table is straightforward: each cell probability must equal the product of its row and column marginal totals.

如果联合分布可分解，则两个随机变量 X 和 Y 独立。对于离散变量，独立性意味着对所有 x, y 有 P(X = x, Y = y) = P(X = x) P(Y = y)。对于连续变量，对所有 x, y 有 f(x, y) = f_X(x) f_Y(y)。独立性意味着知道一个变量的值不能提供关于另一个变量的任何信息。在表格中检验独立性很简单：每个单元格概率必须等于其所在行与列边缘总和的乘积。

In the coin example, does independence hold? 5 ≠ 0.25, so X and Y are not independent. This makes sense because Y is the total heads, which depends on both coins.

在硬币例子中，独立性成立吗？对于 P(X=0, Y=0)=0.25，边缘 P(X=0)=0.5，P(Y=0)=0.25，乘积为 0.125 ≠ 0.25，因此 X 与 Y 不独立。这是合理的，因为 Y 是正面总数，依赖于两枚硬币。

---

7. Expectation, Covariance and Correlation | 期望值、协方差与相关性

The expectation of a function g(X, Y) is E[g(X, Y)] = ΣΣ g(x, y) P(x, y) (discrete) or ∬ Y) are obtained via marginals. Covariance measures the linear association between X and Y:

Cov(X, Y) = E[(X - μ_X)(Y - μ_Y)] = E(XY) - E(X) E(Y)

函数 g(X, Y) 的期望为 E[g(X, Y)] = ΣΣ g(x, y) P(x, y)（离散）或 ∬ g(x, y) f(x, y) dx dy（连续）。特别地，E(X) 和 E(Y) 可通过边缘分布求得。协方差衡量 X 与 Y 之间的线性关系：

Cov(X, Y) = E[(X - μ_X)(Y - μ_Y)] = E(XY) - E(X) E(Y)

If Cov(X, Y) > 0, larger X values tend to pair with larger Y values. If Cov < 0, larger X tends to pair with smaller Y. Covariance strongly depends on the unit of measurement.

如果 Cov(X, Y) > 0，较大的 X 值倾向于与较大的 Y 值配对。如果 Cov < 0，较大的 X 倾向于与较小的 Y 配对。协方差强烈依赖于度量单位。

---

8. The Correlation Coefficient | 相关系数

To obtain a scale-free measure of linear dependence, we use the (Pearson) correlation coefficient:

ρ = Corr(X, Y) = Cov(X, Y) / (σ_X σ_Y),

where σ_X and σ_Y ion; ρ = 0 means no linear correlation (but does not rule out other forms of association).

为了获得一个无量纲的线性相关度量，我们使用（皮尔逊）相关系数：

ρ = Corr(X, Y) = Cov(X, Y) / (σ_X σ_Y),

其中 σ_X 和 σ_Y 是 X 与 Y 的标准差。恒有 -1 ≤ ρ ≤ 1。接近 1 的值表明强正线性相关；接近 -1 表明强负线性相关；ρ = 0 意味着没有线性相关（但不排除其他形式的关联）。

. Remember, independence implies ρ = 0, but ρ = 0 does not necessarily imply independence.

学生必须能够从联合表格或给定的 E(X), E(Y), E(XY), Var(X), Var(Y) 计算 ρ。注意，独立可推出 ρ = 0，但 ρ = 0 不一定推出独立。

---

9. The Bivariate Normal Distribution | 二元正态分布

The bivariate normal distribution is a fundamental continuous model where both X and Y have normal marginals, and their dependence is captured by ρ. Its joint PDF is

f(x, y) = (1 / (2π σ_X σ_Y √1 - ρ² )) exp( -1/(2(1 - ρ²)) [ ((x - μ_X)/σ_X )² - 2ρ ((x - μ_X)/σ_X)((y - μ_Y)/σ_Y) + ((y - μ_Y)/σ_Y)² ] ).

二元正态分布是一个基本的连续模型，其中 X 和 Y 均有正态边缘分布，它们的依赖关系由 ρ 刻画。其联合 PDF 为

f(x, y) = (1 / (2π σ_X σ_Y √1 - ρ² )) exp( -1/(2(1 - ρ²)) [ ((x - μ_X)/σ_X )² - 2ρ ((x - μ_X)/σ_X)((y - μ_Y)/σ_Y) + ((y - μ_Y)/σ_Y)² ] ).

F factorises into two independent normal densities. The bivariate normal distribution is widely used in IB questions to test understanding of conditional expectations, variances, and correlation interpretation.

关键性质：给定 X = x 下 Y 的条件分布也是正态的。若 ρ = 0，联合 PDF 分解为两个独立的正态密度。二元正态分布广泛用于 IB 考题中，以检验对条件期望、方差以及相关性解释的理解。

---

10. Linear Combinations of Bivariate Random Variables | 二元随机变量的线性组合

When we form new random variables such as U = aX + bY, the following rules apply:

• E(aX + bY) = aE(X) + bE(Y), always.
• Var(aX + bY) = a²Var(X) + b²Var(Y) + 2ab Cov(X, Y).

0, and the variance simplifies to a²Var(X) + b²Var(Y). These formulas are essential for solving problems involving total scores, sums, or differences of two related measurements.

当我们构造新的随机变量，如 U = aX + bY 时，适用以下规则：

• E(aX + bY) = aE(X) + bE(Y)，恒成立。
• Var(aX + bY) = a²Var(X) + b²Var(Y) + 2ab Cov(X, Y)。

若 X 与 Y 独立，则 Cov(X, Y) = 0，方差简化为 a²Var(X) + b²Var(Y)。这些公式对于解决涉及总得分、两个相关测量之和或差的问题至关重要。

For the bivariate normal, any linear combination aX + bY also follows a normal distribution. This property is especially useful when constructing confidence intervals or conducting hypothesis tests.

对于二元正态，任何线性组合 aX + bY 也服从正态分布。这一性质在构建置信区间或进行假设检验时尤为有用。

---

11. Common Pitfalls and Exam Tips | 常见错误与考试技巧

Common mistakes:

• Forgetting to check that probabilities sum to 1 when finding unknown parameters.
• Confusing independence with zero correlation.
• Using marginal distributions instead of conditional distributions in prediction problems.
• Incorrectly setting up double integral limits; always sketch the region of support.
• Misapplying the variance formula for sums when variables are correlated.

Exam tips:

• Always list given information: E(X), E(Y), Var(X), Var(Y), Cov, ρ.
• Use the shortcut Cov(X, Y) = E(XY) - E(X)E(Y) unless raw data requires direct computation.
• For discrete tables, calculate marginals first; they are needed for conditionals and independence checks.
• When dealing with bivariate normal, recall that conditional variances are constant and do not depend on the conditioning value.
• Practice past paper questions that blend bivariate normal with transformation of variables.

常见错误：

• 在求未知参数时忘记检查概率总和是否为 1。
• 把独立性与零相关混淆。
• 在预测问题中使用边缘分布而非条件分布。
• 二重积分上下限设置错误；始终画出支撑区域。
• 在变量相关时错误使用和的方差公式。

考试技巧：

• 始终列出给定信息：E(X), E(Y), Var(X), Var(Y), Cov, ρ。
• 除非原始数据需要直接计算，否则使用简便公式 Cov(X, Y) = E(XY) - E(X)E(Y)。
• 对于离散表格，先计算边缘分布；它们在条件分布和独立性检验中不可或缺。
• 处理二元正态时，记住条件方差是常数，不依赖于条件值。
• 练习将二元正态与变量变换相结合的历年真题。

---

Published by TutorHao | Mathematics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）