← 返回资讯屏轩教育 · 行业资讯

Introduction to discrete bivariate distributions | 离散双变量分布简介

2026-08-03 00:45 · 本文部分内容由AI辅助创作

📚 Introduction to discrete bivariate distributions | 离散双变量分布简介

In many real-world situations, we need to study two random variables simultaneously – for example, the height and weight of a person, or the number of study hours and the exam score. When both variables are discrete and we wish to model their joint behaviour, we enter the world of discrete bivariate distributions. This topic forms a natural extension of single-variable probability and lays the groundwork for understanding correlation, regression, and statistical inference.

在许多现实情况中,我们需要同时研究两个随机变量——例如一个人的身高和体重,或者学习小时数与考试成绩。当两个变量都是离散型并且我们希望描述它们共同的行为时,我们就进入了离散双变量分布的世界。本主题是单变量概率的自然延伸,为理解相关性、回归分析和统计推断奠定了基础。

1. What is a Discrete Bivariate Distribution? | 什么是离散双变量分布?

A discrete bivariate distribution describes the probability distribution of a pair of discrete random variables (X, Y). Each variable can only take a countable set of values, such as integers. The distribution tells us how likely each possible combination (x, y) is to occur. For instance, let X be the number of heads when flipping two coins, and Y be the number of tails; then (X, Y) can take pairs like (2,0), (1,1), (0,2). The joint probability function assigns a probability to each pair.

离散双变量分布描述的是一对离散随机变量 (X, Y) 的概率分布。每个变量只能取可数个值,例如整数。该分布告诉我们每一种可能的组合 (x, y) 发生的可能性有多大。例如,设 X 为抛掷两枚硬币时正面朝上的次数,Y 为反面朝上的次数;那么 (X, Y) 可以取 (2,0)、(1,1)、(0,2) 等数对。联合概率函数会为每个数对赋予一个概率。

= x, Y = y) = 1, where P(X = x, Y = y) denotes the joint probability mass function (PMF).

关键要求是所有可能的 (x, y) 对的概率之和为 1,且每个概率介于 0 和 1 之间。用符号表示即 Σₓ Σᵧ P(X = x, Y = y) = 1,其中 P(X = x, Y = y) 表示联合概率质量函数(PMF)。


2. Joint Probability Mass Function | 联合概率质量函数

l (x,y) in the support of (X,Y). For any subset A of the possible pairs, the probability that (X,Y) belongs to A is given by summing f(x,y) over all (x,y) in A.

联合概率质量函数(联合PMF)是离散双变量分布的核心工具。它定义为 f(x,y) = P(X = x, Y = y),其中 (x,y) 属于 (X,Y) 的支撑集。对于可能数对的任意子集 A,(X,Y) 落在 A 中的概率等于对所有 (x,y) ∈ A 的 f(x,y) 求和。

ds (0,1,2), the joint PMF can be tabulated. Each cell entry is the probability of a specific outcome sequence. We must ensure that the assignment of probabilities matches the underlying experiment.

例如,假设我们抛掷一枚均匀硬币两次,定义 X 为第一次抛掷中正面的次数(0 或 1),Y 为总正面次数(0、1、2),则联合PMF可以用表格表示。表中每个单元格是特定结果序列发生的概率。我们必须确保概率的赋值与实际试验相一致。


3. Representing Joint Distributions with Tables | 用表格表示联合分布

A highly convenient way to present a discrete bivariate distribution is via a two-way table. The rows typically correspond to values of X, and the columns to values of Y. Each cell shows the joint probability f(x,y). Consider the following example where X can be 1 or 2, and Y can be 1, 2, or 3:

表示离散双变量分布的一种非常方便的方法是使用双向表。表格的行通常对应 X 的取值,列对应 Y 的取值。每个单元格显示联合概率 f(x,y)。考虑以下例子:X 可取 1 或 2,Y 可取 1、2 或 3:

X \ Y123
10.100.150.05
20.200.300.20

0.30. The sum of all entries equals 1. Tables make it easy to compute further quantities.

从该表格中我们可以直接读出 P(X=2, Y=2) = 0.30。所有表格项之和为 1。表格使得我们能够轻松计算后续的相关量。


4. Marginal Distributions | 边缘分布

variable: P(X = x) = Σᵧ f(x,y). Similarly, the marginal PMF of Y is P(Y = y) = Σₓ f(x,y).

当我们拥有 X 和 Y 的联合分布时,往往也希望单独研究 X 或 Y。单独 X 的概率分布称为 X 的边缘分布。它可以通过对另一变量的所有值求和联合PMF得到:P(X = x) = Σᵧ f(x,y)。类似地,Y 的边缘PMF为 P(Y = y) = Σₓ f(x,y)。

0.05+0.20 = 0.25. These sums are often written in the margins of the table, hence the name 'marginal'.

使用上表,边缘概率 P(X=1) 为行和 0.10+0.15+0.05 = 0.30,P(X=2) = 0.20+0.30+0.20 = 0.70。Y 的边缘分布:P(Y=1) = 0.10+0.20 = 0.30,P(Y=2) = 0.15+0.30 = 0.45,P(Y=3) = 0.05+0.20 = 0.25。这些求和通常写在表格边缘,因此称为“边缘”分布。

Marginal distributions must themselves satisfy the properties of a valid probability mass function: non-negative probabilities summing to 1. They allow us to recover the individual behaviour of each variable ignoring the other.

边缘分布本身必须满足有效概率质量函数的性质:概率非负且总和为 1。它们使我们能够忽略另一个变量而还原每个变量单独的行为。


5. Conditional Distributions | 条件分布

of the other. The conditional probability mass function of Y given X = x is defined as f(y|x) = P(Y = y | X = x), provided P(X = x) > 0. This gives the probability distribution of Y for a fixed x.

有时我们已知其中一个变量已取了某个特定值,需要另一变量的更新分布。给定 X = x 时,Y 的条件概率质量函数定义为 f(y|x) = P(Y = y | X = x) = f(x,y) / P(X = x),前提是 P(X = x) > 0。这给出了固定 x 时 Y 的概率分布。

From the example table, the conditional distribution of Y given X=1 is: P(Y=1|X=1) = 0.10/0.30 = 0.333, P(Y=2|X=1) = 0.15/0.30 = 0.5, P(Y=3| ition on Y values. Conditional distributions must sum to 1 across the conditioned variable's values.

以示例表格为例,给定 X=1 时 Y 的条件分布为:P(Y=1|X=1) = 0.10/0.30 = 0.333,P(Y=2|X=1) = 0.15/0.30 = 0.5,P(Y=3|X=1) = 0.05/0.30 ≈ 0.167。我们也可以对 Y 的值进行条件化。条件分布在被条件变量的所有取值上之和必须为 1。


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

y pair (x,y). Equivalently, the conditional distribution of Y given X equals the marginal distribution of Y.

如果对于所有的 x 和 y,联合PMF 可以分解为边缘PMF 的乘积,即 P(X = x, Y = y) = P(X = x) × P(Y = y),则称离散随机变量 X 和 Y 相互独立。等价地说,给定 X 后 Y 的条件分布等于 Y 的边缘分布。

obability is 0.10, so X and Y are not independent. Independence is a strong condition that implies knowing one variable gives no information about the other. It simplifies many calculations, such as expectations of products.

从之前的表格看,P(X=1) × P(Y=1) = 0.30 × 0.30 = 0.09,而联合概率为 0.10,因此 X 和 Y 不独立。独立性是一个很强的条件,意味着知道一个变量的信息不会给出关于另一个变量的任何信息。它简化了许多计算,例如乘积的期望。


广告 / Sponsored

7. Expectation of Functions of Two Variables | 二元函数的期望

,Y)=X, giving E[X] = Σₓ x P(X=x), which matches the marginal expectation. Similarly for Y.

对于函数 g(X,Y),期望值为 E[g(X,Y)] = Σₓ Σᵧ g(x,y) f(x,y)。这扩展了单变量期望的规则。特别地,X 本身的期望值可取 g(X,Y)=X,得到 E[X] = Σₓ x P(X=x),与边缘期望一致。Y 同理。

res the joint distribution unless X and Y are independent, in which case E[XY] = E[X]E[Y].

一个有用的性质:期望是线性的,因此对于常数 a, b,有 E[aX + bY] = aE[X] + bE[Y]。无论是否独立,该性质均成立。然而对于乘积,E[XY] 通常需要联合分布,除非 X 和 Y 独立,此时 E[XY] = E[X]E[Y]。


8. Covariance | 协方差

] - E[X]E[Y], which is usually simpler. Covariance can be positive (when large X accompanies large Y), negative, or zero. Zero covariance does not necessarily imply independence, but independence implies zero covariance.

协方差衡量 X 与 Y 之间线性关系的强度和方向。它定义为 Cov(X,Y) = E[(X - E[X])(Y - E[Y])]。一个计算公式是 Cov(X,Y) = E[XY] - E[X]E[Y],通常更简单。协方差可以是正的(当较大的 X 伴随较大的 Y 时)、负的、或零。零协方差不一定意味着独立,但独立一定意味着零协方差。

0 = 3.35. Then Cov(X,Y) = 3.35 - 1.70×1.95 = 3.35 - 3.315 = 0.035, a small positive value.

从示例表格可以计算 E[X] = 1×0.30 + 2×0.70 = 1.70,E[Y] = 1×0.30 + 2×0.45 + 3×0.25 = 1.95。 1×3×0.05 + 2×1×0.20 + 2×2×0.30 + 2×3×0.20 = 0.10 + 0.30 + 0.15 + 0.40 + 1.20 + 1.20 = 3.35。于是 Cov(X,Y) = 3.35 - 1.70×1.95 = 3.35 - 3.315 = 0.035,一个很小的正值。


9. Correlation Coefficient | 相关系数

nd σᵧ are the standard deviations of X and Y respectively. The value always lies between -1 and +1. A value of +1 indicates perfect positive linear correlation, -1 perfect negative, and 0 indicates no linear correlation.

相关系数 ρ(或 r)是线性关联的标准化度量,与单位无关。它定义为 ρ = Cov(X,Y) / (σₓ σᵧ),其中 σₓ 和 σᵧ 分别是 X 和 Y 的标准差。其值始终在 -1 和 +1 之间。+1 表示完全正线性相关,-1 表示完全负线性相关,0 表示没有线性相关。

.740. Then ρ = 0.035 / (0.458×0.740) ≈ 0.035 / 0.339 ≈ 0.103, a weak positive correlation.

继续该例子,计算 Var(X) = E[X²] - (E[X])²。E[X²] = 1²×0.30 + 2²×0.70 = 0.30 + 2.80 = 3.10,因此 Var(X) = 3.10 - 1.70² = 3.10 - 2.89 = 0.21,σₓ ≈ 0.458。对于 Y:E[Y²] = 1×0.30 + 4×0.45 + 9×0.25 = 0.30 + 1.80 + 2.25 = 4.35,Var(Y) = 4.35 - 1.95² = 4.35 - 3.8025 = 0.5475,σᵧ ≈ 0.740。于是 ρ = 0.035 / (0.458×0.740) ≈ 0.035 / 0.339 ≈ 0.103,一个弱的正相关。


10. Properties of Covariance and Correlation | 协方差与相关系数的性质

under linear transformations with positive scaling, i.e., ρ(aX+b, cY+d) = sign(ac) ρ(X,Y).

若干重要性质可以简化协方差的计算:Cov(X, X) = Var(X);Cov(X, Y) = Cov(Y, X);对于常数 a,b,c,d,有 Cov(aX + b, cY + d) = ac Cov(X, Y)。特别地,加上常数不影响协方差。相关系数在具有正比例缩放的线性变换下保持不变,即 ρ(aX+b, cY+d) = sign(ac) ρ(X,Y)。

Remember that correlation only captures linear relationships. Two variables can have a perfect quadratic relationship yet zero correlation. Therefore, always pair numerical measures with visual inspection of a scatter plot when possible.

请记住,相关性只捕捉线性关系。两个变量可能具有完美的二次关系而相关性为零。因此,在可能的情况下,务必将数值度量与散点图的视觉检查相结合。


11. Working with More Complex Examples | 处理更复杂的例子

l PMFs; (3) check independence via f(x,y) = fₓ(x) fᵧ(y); (4) compute expectations and variances as needed.

在 IB 风格的题目中,可能会给出一个联合PMF,要求求出使分布有效的常数 k,或者计算 E[XY]、协方差并判断独立性。方法是系统性的:(1) 确保 ΣΣ f(x,y) = 1 以求出未知参数;(2) 计算边缘PMF;(3) 通过 f(x,y) = fₓ(x) fᵧ(y) 检查独立性;(4) 根据需要进行期望和方差的计算。

For instance, suppose f(x,y) = k(2x + y) for x ∈ {1,2} and y ∈ ls and expectations. Practising such step-by-step reasoning is essential for exam success.

例如,假设 f(x,y) = k(2x + y),x ∈ {1,2},y ∈ {1,2}。对所有四种组合求和:k(2+1 + 2+2 + 4+1 + 4+2) = k(3+4+5+6) = 18k = 1,因此 k = 1/18。然后可以继续计算边缘分布和期望。练习这种逐步推理对于考试成功至关重要。


12. Summary and Key Takeaways | 总结与要点

Discrete bivariate distributions allow us to model and analyse the simultaneous behaviour of two discrete random variables. The joint PMF, often displayed in a table, encodes all probabilistic information. From it we derive marginal and conditional distributions, test for independence, and compute measures like covariance and correlation. These concepts not only form a core part of the IB Mathematics curriculum but also serve as stepping stones towards more advanced statistics and probability theory.

离散双变量分布使我们能够建模并分析两个离散随机变量的同时行为。联合PMF 常用表格展示,它包含了所有概率信息。从中我们可以推导边缘分布和条件分布,检验独立性,并计算协方差和相关性等度量。这些概念不仅是 IB 数学课程的核心部分,也是通向更高级统计和概率理论的踏脚石。

Always remember to check that total probability is 1 and that your marginal sums are consistent. With practice, you will be able to handle any discrete bivariate distribution question with confidence and clarity.

请始终记得检查总概率是否为 1,以及边缘和是否一致。通过练习,你将能够自信而清晰地处理任何离散双变量分布问题。

Published by TutorHao | IB Mathematics Revision Series | aleveler.com

广告 / Sponsored
官网正版 国际课程数学核心词汇 唯寻国际教育 全英 学科 GCSE A Level IB MYP IB IBDP 美高…
数学 备考教辅 · 正版现货 · 与本文内容相关
查看详情 / View on Taobao参考价 ¥19.7

更多咨询请联系16621398022(同微信)

备考资讯每日更新,关注不错过
考纲变化 · 科目分析 · 学习方法
扫码关注公众号 tutorhao
在微信里长按二维码识别关注
或截图后用「扫一扫 → 相册」选择本图
公众号:tutorhao