IB数学AI 统计学 正态分布 回归分析 概率 | IB Maths AI Statistics Normal Distribution Regression Analysis Probability

IB数学AI 统计学 正态分布 回归分析 概率

Statistics is arguably the most practical branch of mathematics you will encounter in IB Math Applications & Interpretation. Whether you are analyzing survey data for your IA, interpreting medical trial results, or building predictive models, the statistical toolkit from this course stays with you long after the exam. 统计学可以说是IB数学应用与解释中最实用的分支。无论你是在为IA分析调查数据、解读医学试验结果，还是构建预测模型，这门课程的统计工具箱都会在你考试之后一直陪伴你。 ### 1. What Is Statistics? 什么是统计学？ Statistics is the science of collecting, organizing, analyzing, and drawing conclusions from data. In IB Math AI, statistics is split into two broad branches: descriptive statistics (summarizing and visualizing data) and inferential statistics (making predictions and testing hypotheses). 统计学是收集、整理、分析数据并从中得出结论的科学。在IB数学AI中，统计学分为两大类：描述性统计（总结和可视化数据）和推断性统计（做出预测和检验假设）。 The IB syllabus expects you to move beyond button-pressing on your GDC: you need to interpret what the numbers mean. A p-value of 0.03 is not just "less than 0.05" ： it means there is a 3% probability of observing your data (or more extreme) if the null hypothesis is true. IB大纲希望你不只是会按计算器，而是需要解释数字的含义。p值为0.03不仅仅是"小于0.05"：：它意味着如果原假设为真，观察到你的数据（或更极端数据）的概率只有3%。

Descriptive 描述性

Summarize data with numbers (mean, median, SD) and graphs (histograms, box plots). 用数字和图表总结数据。

Inferential 推断性

Draw conclusions about a population from a sample: confidence intervals, hypothesis tests, chi-squared. 从样本推断总体：置信区间、假设检验、卡方检验。

### 2. Descriptive Statistics 描述性统计 The IB formula booklet gives you the variance formula, but you must know when to use sample standard deviation (s_{n-1}) versus population standard deviation (sigma_n). Rule of thumb: if you have data from a sample and want to estimate the population parameter, use s_{n-1}. If you have the entire population, use sigma_n. IB公式手册给了你方差公式，但你必须知道何时使用样本标准差(s_{n-1})和总体标准差(sigma_n)。经验法则：如果你有样本数据并想估计总体参数，使用s_{n-1}。如果你有整个总体，使用sigma_n。

Measures of Center 集中趋势

Mean: affected by outliers. Median: resistant to outliers. Mode: best for categorical data. 均值受异常值影响，中位数抗异常值，众数最适合分类数据。

Measures of Spread 离散程度

Range: max minus min, crude. IQR: Q3 minus Q1, pairs with box plots. SD: average distance from mean, pairs with normal distribution. 极差粗糙，IQR配箱线图，标准差配正态分布。

The IQR is crucial for identifying outliers. IB exam questions love presenting a data set and asking: "Determine if 87 is an outlier." The rule: any value less than Q1 minus 1.5 times IQR or greater than Q3 plus 1.5 times IQR is an outlier. IQR对于识别异常值至关重要。IB考试题喜欢给出一个数据集然后问："判断87是否为异常值。"规则是：任何小于Q1减去1.5倍IQR或大于Q3加上1.5倍IQR的值都是异常值。 ### 3. Data Presentation 数据展示 Histograms are not bar charts ： this is the single most tested distinction in IB Paper 1. Bar charts show categorical data with gaps between bars; histograms show continuous data with no gaps, and the area of each bar (not height) represents frequency. Use frequency density on the y-axis when bin widths are unequal: frequency density = frequency divided by class width. 直方图不是条形图：：这是IB Paper 1中最常考的区别。条形图展示分类数据，条形之间有间隙；直方图展示连续数据，无间隙，每个条形的面积（而非高度）代表频率。当组距不等时，y轴使用频率密度：频率密度=频率除以组距。

Box Plot 箱线图

Shows 5-number summary: min, Q1, median, Q3, max. Parallel box plots are perfect for comparing two distributions. 展示五数概括：最小值、Q1、中位数、Q3、最大值。平行箱线图非常适合比较两个分布。

Cumulative Frequency 累积频率

Read medians and quartiles directly from the graph. The steeper the curve, the higher the density of data points in that interval. 直接从图上读取中位数和四分位数。曲线越陡，该区间数据点密度越高。

### 4. Bivariate Data and Regression 双变量数据与回归 This is where your GDC earns its keep. You plot a scatter diagram, calculate Pearson's product-moment correlation coefficient r, and if there is a linear trend (r close to 1 or -1), you find the regression line y = ax + b. The key exam pitfall: correlation does NOT imply causation. A strong r-value between ice cream sales and drowning incidents does not mean ice cream causes drowning ： both are driven by summer heat. 这就是你的GDC大显身手的地方。你画出散点图，计算Pearson积矩相关系数r，如果存在线性趋势（r接近1或-1），你求出回归直线y=ax+b。考试关键陷阱：相关性不等于因果性。冰淇淋销量和溺水事件之间的强相关并不意味着冰淇淋导致溺水：：两者都由夏季高温驱动。 For the regression line, IB gives you the formulas for a and b in the formula booklet, but in Paper 2 you will almost always use your GDC. In Paper 1, you may be asked to calculate by hand for small data sets (n = 5 or 6). The regression line always passes through (x-bar, y-bar), the mean point of the data. 对于回归直线，IB在公式手册中给出了a和b的公式，但在Paper 2中你几乎总是使用GDC。在Paper 1中，可能会要求你对小数据集（n=5或6）手动计算。回归直线总是经过数据的均值点(x-bar, y-bar)。 ### 5. Probability Foundations 概率基础 Probability in IB Math AI goes well beyond basic single-event probability. You need conditional probability, independence tests, tree diagrams with multiple branches, and Venn diagram reasoning. The multiplicative law P(A and B) = P(A) times P(B|A) is the workhorse of Paper 1 probability questions. IB数学AI中的概率远不止基本的单一事件概率。你需要条件概率、独立性检验、多分支树状图以及Venn图推理。乘法法则P(A且B)=P(A)乘以P(B|A)是Paper 1概率题的主力。

Key Probability Laws 关键概率法则

Complement Rule 互补法则: P(A') = 1 - P(A)
Addition Rule 加法法则: P(A or B) = P(A) + P(B) - P(A and B)
Conditional Probability 条件概率: P(A|B) = P(A and B) / P(B)
Independence Test 独立性检验: P(A and B) = P(A) times P(B) if and only if A and B are independent
Mutually Exclusive 互斥: P(A and B) = 0 (cannot happen together)

Tree diagrams are your most reliable tool for multi-stage probability. Draw branches with probabilities at each stage, multiply along paths for "and" probabilities, add across paths for "or" probabilities. For conditional probability, highlight the restricted branch and recalculate. 树状图是多阶段概率最可靠的工具。在每个阶段画分支并标注概率，沿路径相乘得到"且"的概率，跨路径相加得到"或"的概率。对于条件概率，高亮受限分支并重新计算。 ### 6. Probability Distributions 概率分布 The binomial distribution models the number of successes in n independent trials, each with probability p of success. Notation: X ~ B(n, p). Your GDC can calculate P(X = k), P(X <= k), and P(a <= X <= b) directly. The mean is np, the variance is np(1-p). 二项分布模型：在n次独立试验中成功的次数，每次成功概率为p。记号：X~B(n,p)。你的GDC可以直接计算P(X=k)、P(X<=k)和P(a<=X<=b)。均值为np，方差为np(1-p)。 The normal distribution is the centerpiece of IB statistics. Every normal curve is determined by its mean mu and standard deviation sigma. The empirical rule: approximately 68% of data lies within 1 sigma of mu, 95% within 2 sigma, and 99.7% within 3 sigma. To use the z-score table, standardize: z = (x - mu) / sigma. Then find the probability from the table or your GDC. 正态分布是IB统计学的核心。每条正态曲线由其均值mu和标准差sigma决定。经验法则：大约68%的数据落在mu的正负1个sigma之内，95%在2个sigma之内，99.7%在3个sigma之内。使用z分数表时进行标准化：z=(x-mu)/sigma。然后从表格或GDC中找到概率。

Binomial 二项分布

Count successes. Fixed n, constant p, independent trials. Use when counting "how many pass the test." 计数成功次数。固定n，恒定p，独立试验。用于"多少人通过测试"类问题。

Normal 正态分布

Continuous, bell-shaped. Defined by mu and sigma. Use for heights, weights, test scores, measurement errors. 连续型，钟形曲线。由mu和sigma定义。用于身高、体重、考试分数、测量误差。

A classic IB question: "The masses of apples are normally distributed with mean 150 g and standard deviation 12 g. Find the probability that a randomly chosen apple has a mass between 140 g and 165 g." z1 = (140-150)/12 = -0.833, z2 = (165-150)/12 = 1.25. Then P(-0.833 < Z < 1.25) = P(Z < 1.25) - P(Z < -0.833) = 0.8944 - 0.2023 = 0.6921. 经典IB题目："苹果质量的均值为150克，标准差为12克，服从正态分布。求随机选一个苹果质量在140克到165克之间的概率。" ### 7. Hypothesis Testing 假设检验 Hypothesis testing is where statistics becomes a decision-making tool. The structure is always the same: state H0 (null) and H1 (alternative), choose significance level alpha (usually 5%), calculate the test statistic or p-value, and compare. If p-value < alpha, reject H0. If p-value >= alpha, do not reject H0. Never say "accept H0" ： you can only fail to reject it. 假设检验是统计学成为决策工具的地方。结构总是相同的：陈述H0（原假设）和H1（备择假设），选择显著性水平alpha（通常5%），计算检验统计量或p值，然后比较。如果p值=alpha，不拒绝H0。永远不要说"接受H0"：：你只能无法拒绝它。

One-Tailed vs Two-Tailed 单尾与双尾

One-tailed 单尾: H1 has ">" or "<". All of alpha goes into one tail. Use when you have a directional prediction. Example: "The new drug reduces blood pressure." 有方向性预测时使用。

Two-tailed 双尾: H1 has "not equal". Alpha is split equally between both tails. Use when you are only testing for a difference, not a direction. Example: "The new drug changes blood pressure." 仅测试是否有差异时使用。

The most common IB hypothesis test is the t-test for a population mean when sigma is unknown. You calculate t = (x-bar - mu) / (s / sqrt(n)) and compare against the critical value from the t-distribution with n-1 degrees of freedom. Your GDC has a dedicated t-test function ： learn it well. 最常见的IB假设检验是sigma未知时对总体均值的t检验。你计算t=(x-bar-mu)/(s/sqrt(n))并与自由度为n-1的t分布临界值比较。你的GDC有专门的t检验功能：：好好掌握它。 ### 8. Chi-Squared Tests 卡方检验 The chi-squared test assesses whether observed frequencies differ significantly from expected frequencies. There are two flavors in the IB syllabus: the goodness-of-fit test (one categorical variable: does this die roll fairly?) and the test for independence (two categorical variables: is voting preference independent of age group?). 卡方检验评估观察频率是否与期望频率有显著差异。IB大纲中有两种类型：拟合优度检验（一个分类变量：这个骰子是否公正？）和独立性检验（两个分类变量：投票偏好是否独立于年龄组？）。 The test statistic is chi-squared = sum((O - E)^2 / E) where O is the observed frequency and E is the expected frequency. Degrees of freedom: for goodness of fit, df = number of categories minus 1; for independence, df = (rows minus 1) times (columns minus 1). All expected frequencies must be at least 5 for the test to be valid. 检验统计量为卡方=sum((O-E)^2/E)，其中O是观察频率，E是期望频率。自由度：拟合优度df=类别数减1；独立性df=(行数减1)乘以(列数减1)。所有期望频率必须至少为5才有效。

Chi-Squared Worked Example 卡方计算示例

A die is rolled 120 times. Observed: 15, 22, 18, 25, 19, 21. Expected each: 20. chi-squared = (15-20)²/20 + (22-20)²/20 + (18-20)²/20 + (25-20)²/20 + (19-20)²/20 + (21-20)²/20 = 1.25 + 0.20 + 0.20 + 1.25 + 0.05 + 0.05 = 3.00. df = 6 - 1 = 5. Critical value at 5%: 11.07. Since 3.00 < 11.07, we do not reject H0: the die appears fair. 不拒绝H0：骰子看起来是公正的。

### 9. IB Exam Tips IB考试技巧 **Paper 1 (no calculator)** IB frequently asks you to read values from cumulative frequency graphs, interpret box plots, and perform simple probability calculations. Memorize the normal distribution symmetry: P(Z < -a) = P(Z > a) = 1 - P(Z < a). You do not need to memorize the full z-table ： IB provides relevant values if needed. IB经常要求你从累积频率图中读取值、解读箱线图并进行简单的概率计算。记住正态分布的对称性：P(Z<-a)=P(Z>a)=1-P(Z<a)。你不需要记住完整的z表：：IB会在需要时提供相关值。 **Paper 2 (calculator required)** Your GDC is your best friend. Know exactly where to find: normalcdf, invNorm, binompdf, binomcdf, t-test, chi-squared test, and linear regression. Practice the GDC menu navigation until it is muscle memory ： losing 30 seconds per function call adds up across 10 questions. 你的GDC是你最好的朋友。确切知道在哪里找到：normalcdf、invNorm、binompdf、binomcdf、t检验、卡方检验和线性回归。练习GDC菜单导航直到成为肌肉记忆：：每次函数调用浪费30秒，10道题下来累积可观。 **IA connection** Statistics topics are the most popular choice for IB Math AI Internal Assessments. A well-executed chi-squared test or regression analysis with real data makes a strong IA. Choose a topic you genuinely care about ： the examiner can tell the difference between genuine curiosity and forced data hunting. 统计学主题是IB数学AI内部评估最受欢迎的选择。一个有真实数据支撑且执行良好的卡方检验或回归分析能做出优秀的IA。选择一个你真正关心的主题：：考官能分辨出真正的求知欲和被迫寻找数据的区别。 ### 10. Calculator Guide 计算器使用指南 For IB Math AI, the Casio fx-991CN X is the go-to calculator. It handles normal distribution probabilities, binomial calculations, t-tests, and chi-squared tests natively. The key GDC functions you must master: MENU 6 (Statistics) for entering data and computing mean, SD, and regression; MENU 7 (Distribution) for normal/binomial probabilities; and the Chi-squared GOF test under the Statistics menu. 对于IB数学AI，Casio fx-991CN X是首选计算器。它原生支持正态分布概率、二项分布计算、t检验和卡方检验。你必须掌握的关键GDC功能：MENU 6（统计）用于输入数据和计算均值、标准差和回归；MENU 7（分布）用于正态/二项概率；以及统计菜单下的卡方拟合优度检验。 One subtlety: when you run a t-test on your GDC, pay attention to whether you select "1-Sample" (compare one sample to a hypothesized mean) or "2-Sample" (compare two independent samples). This is the most common GDC menu mistake IB students make in the exam. 一个细微之处：在GDC上运行t检验时，注意选择"1-Sample"（将一个样本与假设均值比较）还是"2-Sample"（比较两个独立样本）。这是IB学生在考试中最常见的GDC菜单错误。 ### 11. Common Mistakes to Avoid 避坑指南 1

Confusing sample and population SD: Use s_{n-1} for samples, sigma_n for populations. Exams penalize this heavily. 混淆样本和总体标准差：样本用s_{n-1}，总体用sigma_n。考试对此扣分严厉。

2

Saying "accept H0": You never accept the null hypothesis. Correct phrasing: "There is insufficient evidence to reject H0." 永远不要说"接受原假设"。正确表述："没有足够证据拒绝H0。"

3

Violating chi-squared conditions: All expected frequencies must be >= 5. If not, merge categories or use a different test. 违反卡方条件：所有期望频率必须>=5。如不满足，合并类别或使用其他检验。

4

Using height instead of area for histograms: When class widths are unequal, the bar height is frequency density, not frequency. IB designs questions specifically to trap students on this. 直方图用高度而非面积：组距不等时，条形高度是频率密度而非频率。IB专门设计题目来让学生掉进这个陷阱。

5

Using normal approximation for binomial without checking np and n(1-p): Both must be >= 10 for the approximation to be valid. 使用二项的正态近似前不检查np和n(1-p)：两者必须都>=10才有效。

6

Correlation = causation: The most cited statistical fallacy. Always address lurking variables and confounding factors in your IA and exam answers. 相关等于因果：最常被引用的统计谬误。在IA和考试答案中始终处理潜伏变量和混杂因素。

### 12. Study Plan: Quick Revision Guide 快速复习指南

Day 1-2 第一天至第二天

Descriptive stats + data presentation. Practice reading box plots and cumulative frequency graphs. 描述统计+数据展示。练习解读箱线图和累积频率图。

Day 3-4 第三天至第四天

Bivariate data + regression. Do 5 scatter plot questions with your GDC. 双变量数据+回归。用GDC做5道散点图题。

Day 5-6 第五天至第六天

Probability + distributions. Master tree diagrams and normal/binomial GDC functions. 概率+分布。掌握树状图和正态/二项GDC函数。

Day 7 第七天

Hypothesis testing + chi-squared. Write full test conclusions for 5 past paper questions. 假设检验+卡方。为5道历年真题写完整的检验结论。

A practical tip: for each past paper question you attempt, time yourself. IB gives roughly 1 minute per mark. If a 6-mark statistics question takes you 12 minutes, you are spending twice as long as you should ： either your GDC navigation needs practice or your conceptual understanding needs reinforcement. 实用建议：对于每道你做的历年真题，计时完成。IB大约每分给1分钟时间。如果一道6分的统计题花了你12分钟，你花了两倍的时间：：要么你的GDC操作需要练习，要么你的概念理解需要加强。 Practice writing hypothesis test conclusions in full sentences. IB examiners award a mark specifically for the contextualized conclusion: "There is sufficient evidence at the 5% significance level to conclude that the mean mass of apples from Orchard B is greater than 150 g." Generic answers like "reject H0" lose this mark. 练习用完整句子写假设检验结论。IB考官会专门为结合上下文的结论给分："在5%显著性水平上有足够证据推断果园B的苹果平均质量大于150克。"像"拒绝H0"这样的泛泛答案会丢掉这一分。

🛍️ 微信小店好物推荐

fx-991CN X 中文版计算器-老王推荐

¥168.00 🛒 微信小店购买

🎓 Need one-on-one tutoring? 需要一对一辅导?

📱 16621398022 同微信

Follow tutorhao on WeChat for more learning resources 关注公众号获取更多学习资源

更多咨询请联系16621398022（同微信）