Model Selection Methods in Statistical Modelling | 统计建模中的模型选择方法

📚 Model Selection Methods in Statistical Modelling | 统计建模中的模型选择方法

Statistical modelling is central to the A-Level Further Mathematics syllabus, especially in Edexcel Further Statistics. Model selection is the process of choosing the best model, balancing goodness of fit with complexity. This article explores the main methods you need for your exams.

统计建模是 A-Level 进阶数学课程的核心内容，尤其在 Edexcel 进阶统计中。模型选择就是在拟合优度与模型复杂度之间寻找平衡的过程。本文围绕考试要求，系统讲解模型选择的常用方法。

1. Why Model Selection Matters | 模型选择的意义

In statistics, a model is a simplified description of a real-world process. A good model captures the essential pattern without chasing random noise. Model selection is the set of rules we use to decide which model to keep.

在统计学中，模型是对现实过程的简化描述。好的模型应捕捉核心规律，而不是追逐随机噪声。模型选择就是帮助我们决定保留哪个模型的一套规则。

For example, in weather forecasting, a very simple model may miss an important temperature trend, while an over-complicated model may give unreliable predictions for tomorrow. Choosing the right level of detail is the task of model selection.

例如，在天气预报中，过于简单的模型可能错过重要的气温趋势，而过于复杂的模型可能对明天的天气给出不可靠的预测。选择恰当的复杂程度正是模型选择的任务。

For Edexcel Further Statistics, you need to know how to compare models using hypothesis tests, information criteria and diagnostic checks. These tools help you answer questions such as “Is a linear or quadratic regression better?” or “Should we include this variable?”

对于 Edexcel 进阶统计，你需要掌握如何借助假设检验、信息准则和诊断检验比较模型。这些工具可以回答类似 “线性回归还是二次回归更好？” 或 “该不该加入这个变量？” 的问题。

2. Overfitting and Underfitting | 过拟合与欠拟合

Underfitting occurs when a model is too simple to capture the underlying trend. Overfitting occurs when a model is too complex and fits the sample data too closely, including random errors.

欠拟合是指模型过于简单，无法捕捉潜在趋势；过拟合则

Published by TutorHao | A-Level 进阶数学 Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）