📚 IB Maths: 13F Modelling Skills | IB数学：13F建模技能
Modelling is the art of translating real-world phenomena into mathematical language. In IB Mathematics, particularly in the Applications and Interpretation (AI) course, modelling skills are essential. You will be expected to choose appropriate functions, fit them to data, and use the model to make predictions, all while evaluating its limitations. This article explores the key modelling skills covered in topic 13F.
建模是将现实世界现象转化为数学语言的艺术。在IB数学课程，尤其是应用与解释（AI）课程中，建模技能至关重要。你需要选择恰当的函数，将其拟合到数据中，并利用模型进行预测，同时评估其局限性。本文探讨主题13F所涵盖的核心建模技能。
1. What is Mathematical Modelling? | 什么是数学建模？
Mathematical modelling involves creating a mathematical representation of a real-world situation. It could be a function that describes population growth, the trajectory of a ball, or the cooling of a cup of coffee. The goal is to understand the system, make predictions, and sometimes optimise outcomes. Unlike pure 'textbook' problems, modelling problems contain noise, limited data, and require you to decide which variables matter most.
数学建模涉及创建现实世界情境的数学表示。它可以是一个描述人口增长的函数，一个球的轨迹，或一杯咖啡的冷却。目标是理解系统，做出预测，有时优化结果。与纯粹的“教科书”问题不同，建模问题包含噪声、有限的数据，并要求你决定哪些变量最为重要。
2. The Modelling Cycle | 建模循环
s. A typical flow is: Real-world problem → Assumptions → Mathematical model → Mathematical solution → Interpretation → Validation → Refined model.
建模过程是循环的：从现实问题开始，简化假设，建立数学模型，进行数学分析，将结果解释回现实，与数据验证，如有必要，改进模型。这一循环是IB内部评估的核心。典型的流程是：现实问题 → 假设 → 数学模型 → 数学解 → 解释 → 验证 → 改进模型。
3. Choosing a Model | 选择合适的模型
The first step in building a model is to recognise patterns in the data. Plotting a scatter diagram helps decide which family of functions to use. If the points roughly lie on a straight line, choose a linear model. If they show a single curve with no reversal, a quadratic or exponential model may fit. Regular oscillations suggest a sinusoidal model. Always consider the context: a population growing without limit might be exponential, while a ball's height follows a quadratic path.
构建模型的第一步是识别数据中的模式。绘制散点图有助于决定使用哪类函数。如果点大致落在一条直线上，选择线性模型。如果它们呈现单一弯曲而无反转，二次或指数模型可能合适。规律的振荡表明应使用正弦模型。始终要考虑情境：无限增长的人口可能是指数型的，而球的高度则遵循二次路径。
4. Linear Models | 线性模型
-intercept (initial value when x = 0). Use this when a constant rate of change is expected – ated by picking two points on the line of best fit or by using linear regression on a GDC.
线性模型是最简单的：y = mx + c，其中 m 是斜率（变化率），c 是 y 截距（当 x = 0 时的初始值）。当预期存在恒定变化率时使用，例如随着重量增加弹簧的长度，或简单的成本分析。参数 m 和 c 可以通过在最佳拟合线上取两点或用GDC进行线性回归来估计。
y = mx + c
5. Quadratic Models | 二次模型
k (vertex form). They describe situations involving projectile motion, area optimisation, or any symmetrical rise and fall. The sign of a determines whether the parabola opens up (a > 0) or down (a < 0). The vertex gives the maximum or minimum point, often the most critical feature in a modelling context.
二次模型的形式为 y = ax² + bx + c（一般式）或 y = a(x − h)² + k（顶点式）。它们描述涉及抛射体运动、面积优化或任何对称升降的情境。a 的符号决定抛物线是开口向上（a > 0）还是向下（a < 0）。顶点给出最大值或最小值点，这通常是建模情境中最关键的特征。
y = ax² + bx + c
6. Exponential Models | 指数模型
r decay that changes by a constant percentage per unit time. The general forms are y = a·bˣ or y = a·eᵏˣ population growth, radioactive decay, and compound interest. Note that exponential functions never reach zero – they approach an asymptote.
指数模型非常适合每单位时间以恒定百分比变化的增长或衰减。一般形式为 y = a·bˣ 或 y = a·eᵏˣ，其中 a 是初始量，b 是增长因子（b > 1 表示增长，0 < b < 1 表示衰减），k 是连续增长率。常见例子包括人口增长、放射性衰变和复利。注意指数函数永远不会达到零——它们趋近于一条渐近线。
y = a·eᵏˣ
7. Sinusoidal Models | 正弦模型
, and d is the vertical shift (the principal axis). When fitting a sine curve, identify the maximum and minimum values to find amplitude and vertical shift, then estimate the period from the data.
正弦模型用于规律振荡的数据，如潮汐、声波或季节性温度变化。一般形式为 y = a sin(b(x − c)) + d，其中 a 是振幅，b 决定周期（周期 = 2π/b），c 是水平位移，d 是垂直位移（主轴线）。拟合正弦曲线时，先确定最大值和最小值以求出振幅和垂直位移，再从数据中估计周期。
y = a sin(b(x − c)) + d
8. Fitting Parameters to Data | 拟合参数
g, QuadReg, ExpReg, SinReg) quickly produce optimal coefficients. Always store the regression equation in your Y= editor to plot it against the scatter plot and visually check the fit.
一旦选定模型类型，下一步是确定参数（例如 m、c、a、b、k）。在IB考试中，你可能要用两点手动求线性模型，但更常见的是使用GDC进行回归。对于二次和指数，GDC回归函数（LinReg、QuadReg、ExpReg、SinReg）可快速得到最优系数。务必将回归方程存储在 Y= 编辑器中，将其与散点图一起绘制，目视检查拟合情况。
9. Assessing the Model: Residuals and R² | 模型评估：残差和R²
erences between observed data (y-data) and the values predicted by the model: residual = yactual − y_predicted. A residual plot should show no clear pattern – o 1 suggests a very strong fit. IB exam questions often ask you to comment on the suitability of a model based on these indicators.
一个好模型不仅仅是得到一个方程；你必须评估它的拟合程度。残差是观测数据（y数据）与模型预测值之间的垂直差异：残差 = y_实际 − y_预测。残差图应显示无明显模式——随机性表示拟合良好。决定系数 R² 衡量模型解释的方差比例：接近 1 的值表示拟合非常强。IB考试题经常要求你根据这些指标评价模型的适用性。
10. Refining the Model | 改进模型
Even with a high R², a model may fail at the extremes or miss underlying behaviour. Refining can involve using a piecewise function – different formulas for different intervals – or adding an extra term. For instance, a quadratic might model an object's height well over short times, but a logistic function would be better for long-term population with a carrying capacity. However, beware of overfitting: adding too many parameters can make the model fit the current data perfectly but perform poorly on new data.
即使 R² 很高，模型在极端情况下也可能失效或忽略了潜在行为。改进可以包括使用分段函数——在不同区间使用不同公式——或增加额外项。例如，二次函数可能很好地模拟物体短时间的高度，但对于有承载力的长期人口，逻辑函数会更好。但要警惕过拟合：增加过多参数可能使模型完美拟合当前数据，但对新数据表现极差。
11. Technology in Modelling | 建模中的技术应用
Graphic display calculators (GDC) are indispensable in IB modelling. Beyond regression, use your GDC to plot data, adjust window settings, and trace points. For those using software like GeoGebra or Desmos, sliders allow you to manually adjust parameters and observe the curve in real time, deepening your understanding of how each coefficient changes the shape. In the exam, knowing the exact keystrokes for regression and storing equations saves valuable time.
图形显示器计算器（GDC）在IB建模中不可或缺。除了回归，使用GDC绘制数据、调整窗口设置和追踪点。对于使用GeoGebra或Desmos等软件的同学，滑动条允许手动调整参数并实时观察曲线，加深对每个系数如何改变形状的理解。在考试中，熟练操作回归并存储方程的按键能节约宝贵时间。
12. Common Pitfalls and Exam Tips | 常见错误与考试技巧
Always check the domain of your model – a linear model for ice cream sales might not apply when the temperature drops below 0°C. State units clearly with any value. When predicting, distinguish between interpolation (within the data range) and extrapolation (outside): extrapolation is risky and requires caution. In an exam, if asked 'to what extent this model is suitable', always refer to the residual plot and R², and mention any contextual limitations. Finally, link your conclusion back to the original real-world problem – modelling is never just about numbers.
务必检查模型的定义域——冰淇淋销售的线性模型在温度降到0°C以下时可能不适用。任何数值都要清晰标注单位。预测时，区分内插（数据范围内）和外推（范围外）：外推有风险，需要谨慎。在考试中，如果被问及“此模型在多大程度上合适”，务必提及残差图和 R²，并指出情境限制。最后，将结论联系回原始现实问题——建模从来不只是关于数字。
Published by TutorHao | IB Maths Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图