The Large Data Set | 大型数据集

📚 The Large Data Set | 大型数据集

The Large Data Set (LDS) is a real‑world collection of meteorological records that forms an integral part of the Edexcel A‑Level Mathematics specification. Working with this data set allows you to practise the full statistical cycle: posing questions, sampling, cleaning data, calculating summary statistics, drawing graphs and drawing conclusions in context. Mastery of the LDS is essential because examination questions often provide extracts from it and expect confident manipulation, interpretation and comparison of the figures.

大型数据集（LDS）是一组真实的气象记录，是Edexcel A‑Level数学大纲的核心内容。处理这个数据集让你有机会完整地实践统计循环——提出问题、进行抽样、清理数据、计算汇总统计量、绘制图形并在实际背景中得出结论。精通大型数据集至关重要，因为考试题目常常提供其中的摘录，并期望考生能够熟练地操作、解读以及比较这些数据。

1. Introduction to the Large Data Set | 大型数据集简介

The Edexcel Large Data Set is not just a random spreadsheet; it is a carefully selected set of daily weather measurements from eight locations across the world, covering specific months in 1987 and 2015. By embedding this authentic data into the course, the exam board ensures that students experience genuine statistical investigation — from handling missing values and anomalies to making valid comparisons between different climates.

Edexcel 的大型数据集并不是一份随机的电子表格，而是一组精心选取的全球八个地点的每日天气测量数据，覆盖了 1987 年和 2015 年的特定月份。通过将这组真实数据纳入课程，考试局确保学生能够经历真正的统计探究过程——从处理缺失值和异常点到对不同气候进行有效比较。

The data set is provided as a spreadsheet and can be filtered, sorted and sampled. In the examination, you may be given a printed excerpt or a summary table, and you must demonstrate that you understand the structure of the variables, the units of measurement and the potential pitfalls when drawing conclusions.

该数据集以电子表格的形式提供，可以进行筛选、排序和抽样。在考试中，你可能会拿到一份印刷摘录或汇总表格，你必须证明自己理解变量的结构、测量单位以及在得出结论时可能遇到的陷阱。

2. Structure and Variables of the LDS | 大型数据集的结构与变量

The LDS contains records for five UK weather stations (Leuchars, Leeming, Heathrow, Hurn and Camborne) and three overseas stations (Jacksonville, Beijing and Perth). For each location and date, the following variables are recorded:

大型数据集包含五个英国气象站（Leuchars、Leeming、Heathrow、Hurn 和 Camborne）以及三个海外气象站（Jacksonville、Beijing 和 Perth）的记录。每个地点和日期都记录了下列变量：

Daily Mean Temperature (°C)

日均气温（摄氏度）
Daily Total Rainfall (mm)

日总降雨量（毫米）
Daily Total Sunshine (hours)

日总日照时数（小时）
Daily Mean Wind Direction (degrees)

日均风向（度）
Daily Mean Wind Speed (knots)

日均风速（节）
Daily Maximum Gust (knots)

日最大阵风（节）
Cloud Cover (oktas)

云量（八分之几）
Visibility (metres)

能见度（米）
Pressure (hPa)

气压（百帕）

All these variables, together with the date and the location identifier, form a rectangular data set where each row represents one day’s observations. Knowing the units and the typical range of each variable is critical for spotting outliers and interpreting summary statistics correctly.

所有这些变量，再加上日期和地点标识，构成了一个矩形数据集，其中每一行代表一天的观测结果。了解每个变量的单位和典型范围对于发现异常值以及正确解读汇总统计量至关重要。

3. Types of Data in the LDS | 大型数据集中的数据类型

Most of the variables in the LDS are continuous: temperature, rainfall, sunshine duration, wind speed, gust speed, visibility and pressure can all take any value within a given interval. Cloud cover, measured in oktas, is discrete because it can only take the integer values 0, 1, 2, …, 8. Wind direction is measured in degrees and is therefore also continuous, but it is often treated as a circular variable. The date and location are categorical (or ordinal in the case of date) and are used for grouping and filtering.

大型数据集中的大部分变量是连续型的：温度、降雨量、日照时数、风速、阵风速度、能见度和气压都可以在给定区间内取任意值。云量以八分之几为单位，是离散型的，因为它只能取整数值 0、1、2、…、8。风向以度为单位，因此也是连续型的，但常被视作圆形变量。日期和地点是分类变量（日期具有顺序性），用于分组和筛选。

Recognising the type of data helps you choose appropriate diagrams and calculations. For continuous data you might draw histograms or box plots; for discrete cloud cover a bar chart could be suitable. In exam tasks, you must identify whether a variable is continuous, discrete or categorical to justify your statistical choices.

识别数据类型有助于选择合适的图表和计算方式。对于连续型数据，你可以绘制直方图或箱线图；对于离散的云量，条形图可能较为合适。在考试任务中，你必须判断变量是连续型、离散型还是分类型，从而为你的统计选择提供理由。

4. Sampling from the Large Data Set | 从大型数据集中抽样

The complete LDS contains thousands of rows, so in an examination you will often work with a sample. You need to understand how simple random sampling, systematic sampling, stratified sampling and opportunity sampling could be applied to this data. For instance, you could take a simple random sample of 50 days from Heathrow, or use stratified sampling to ensure proportional representation of the five UK sites.

完整的大型数据集包含数千行数据，因此在考试中你通常需要处理一个样本。你需要理解如何将简单随机抽样、系统抽样、分层抽样和机会抽样应用到该数据上。例如，你可以从希斯罗机场的数据中抽取一个包含 50 天的简单随机样本，或者采用分层抽样来确保五个英国站点按比例被代表。

A common examination task is to describe how you would select a sample from the LDS. You must mention practical steps such as numbering the rows, using a random number generator, and deciding whether to include or exclude missing values. Explaining the advantages of one method over another — for example, that stratified sampling gives more representative sub‑groups — demonstrates deeper understanding.

常见的考试任务是描述你如何从大型数据集中选取样本。你必须提到实际步骤，例如为行编号、使用随机数生成器，以及决定是否包含或排除缺失值。解释某种方法相较于另一种方法的优势——例如分层抽样能提供更具代表性的子组——可以体现出更深入的理解。

5. Cleaning the Data: Identifying Errors and Anomalies | 数据清理：识别错误与异常

Real data is rarely perfect, and the LDS contains entries that require cleaning. Some values are recorded as ‘n/a’ (not available) or ‘tr’ (trace, for rainfall less than 0.05 mm). Before calculating statistics, you must decide how to handle these — either by removing the day entirely or by replacing the symbol with a numerical value when instructed. You may also find obvious errors, such as a wind speed of 999 knots or a negative rainfall reading.

真实数据很少是完美无缺的，大型数据集中就包含一些需要清理的记录。有些数值被记录为 ‘n/a’（不可用）或 ‘tr’（微量，表示降雨量小于 0.05 mm）。在进行统计计算之前，你必须决定如何处理这些值——根据题目要求，要么将整个该日数据删除，要么用一个数值替换该类符号。你还可能发现明显的错误，例如风速为 999 节，或降雨量为负值。

Anomalies are values that are genuinely possible but highly unusual — for example, a gust of 60 knots at an inland station. Cleaning involves both removing impossible values and noting unusual ones that might be outliers. In the exam you might be asked to suggest a reason for an outlier, such as a storm event, and to comment on whether it should be kept for a realistic picture of the weather.

异常值是指那些确实可能发生，但极不寻常的数值——例如，一个内陆测站出现了 60 节的阵风。数据清理既包括剔除不可能出现的数值，也包括标记那些可能为离群值的罕见数据。在考试中，你可能会被问及对某个异常值给出可能的解释，例如一次风暴事件，并评价是否应当保留该值以呈现真实的天气状况。

6. Measures of Central Tendency (Location) | 集中趋势的度量（位置）

When summarising a variable from the LDS, you will calculate the mean, median and mode. The mean daily temperature at Heathrow in July 2015 could be found by summing all available temperatures and dividing by the number of days: x̄ = Σx/n. The median is the middle value when the data are ordered, which is robust to extreme values; on a day with an unusually low temperature, the median would stay representative.

在汇总大型数据集中的某一变量时，你需要计算均值、中位数和众数。例如，希斯罗机场 2015 年 7 月的日均温度可以通过将所有可得温度相加并除以天数而求得：x̄ = Σx/n。中位数是将数据排序后处于中间位置的数值，它对极端值具有稳健性；如果某天温度异常低，中位数依然能保持代表性。

For discrete variables like cloud cover, the mode can be useful. You might find that an aviation station such as Leuchars has a modal cloud cover of 8 oktas in winter, indicating predominantly overcast skies. Exam questions often ask you to interpret what the mean and median reveal about the shape of a distribution — if the mean is noticeably higher than the median, the data are likely positively skewed.

对于云量这类离散型变量，众数可能很有用处。你或许会发现像 Leuchars 这样的航空测站在冬季的云量众数为 8 oktas，表明天空以阴天为主。考试题目常常要求你解释均值和中位数揭示了分布形状的哪些信息——如果均值明显高于中位数，数据很可能呈正偏态。

7. Measures of Dispersion (Spread) | 离散程度的度量（离散）

Alongside an average, you must quantify variability. The range (max − min) is simple but sensitive to outliers. For the LDS, the interquartile range (IQR = Q₃ − Q₁) gives a more resistant measure of spread. The sample variance s² = Σ(x − x̄)²/(n−1) and standard deviation s = √s² summarise how tightly the data cluster around the mean.

除了平均数之外，你还必须量化变异性。极差（最大值 − 最小值）计算简单，但对异常值很敏感。对于大型数据集，四分位距（IQR = Q₃ − Q₁）能够提供更稳健的离散度量。样本方差 s² = Σ(x − x̄)²/(n−1) 和标准差 s = √s² 则概括了数据围绕均值的集中程度。

In an exam, you may be given summary statistics from two stations and asked to compare consistency. For instance, Camborne might have a lower standard deviation for daily total sunshine in August than Heathrow, indicating more reliable sunny weather. Remember to use correct units when quoting spread: if rainfall is measured in mm, the standard deviation is also in mm.

在考试中，你可能会被给出两个测站的汇总统计量，并被要求比较其稳定性。例如，Camborne 八月份的日总日照时数的标准差可能低于希斯罗，表明那里的晴朗天气更加稳定。请记住，在引用离散程度时要使用正确的单位：如果降雨量以 mm 为单位，那么标准差的单位也是 mm。

8. Graphical Representations of the Data | 数据的图形表示

The LDS lends itself to a variety of statistical diagrams. Box plots allow you to compare the five‑number summaries (minimum, Q₁, median, Q₃, maximum) of temperature or pressure at different locations. Histograms are ideal for showing the distribution of continuous variables like daily total sunshine, and you must be able to calculate frequency density and interpret area as proportional to frequency.

大型数据集适合绘制多种统计图形。箱线图可以让你比较不同地点的温度或气压的五数概括（最小值、Q₁、中位数、Q₃、最大值）。直方图非常适合展示日总日照时数这类连续型变量的分布情况，你必须能够计算频率密度，并将面积理解为与频率成正比。

Cumulative frequency curves provide a way to estimate medians, quartiles and percentiles directly from grouped data. Scatter diagrams can be used to explore potential relationships, for example plotting daily maximum gust against daily mean wind speed to see if higher average winds are associated with stronger gusts. Always label axes clearly, include units and give the chart a meaningful title.

累积频率曲线提供了一种直接从分组数据中估计中位数、四分位数和百分位数的方法。散点图可用于探索潜在的关系，例如，以日最大阵风对日均风速作图，观察较高的平均风速是否与较强的阵风相关联。务必清晰地标注坐标轴、注明单位，并为图表设置一个有意义的标题。

9. Comparing Data Sets Using LDS | 使用大型数据集进行比较

Edexcel often asks you to compare distributions using summary statistics from two locations or two time periods. A strong comparison goes beyond stating ‘Perth is hotter than Beijing’; you should use values of median and IQR to describe both the central tendency and the consistency. You might write: ‘The median daily mean temperature in July at Perth (18.2°C) is higher than at Beijing (14.8°C), and the IQR is smaller (3.1°C compared with 5.4°C), suggesting that Perth’s temperatures are not only warmer but also less variable.’

Edexcel 经常要求你利用两个地点或两个时段的汇总统计量来比较分布。有力的比较绝不会仅仅停留在“珀斯比北京热”这样的表述上；你应当使用中位数和四分位距的数值来描述集中趋势和稳定性。你可能会这样写：“珀斯七月份的日均气温中位数（18.2°C）高于北京（14.8°C），且四分位距更小（3.1°C 对比 5.4°C），这表明珀斯的气温不仅更暖和，而且变化也更小。”

Skewness should also be addressed through comparison. If one distribution has a mean much larger than its median while the other is roughly symmetric, you can conclude that the first is positively skewed. Mentioning real‑weather reasons — such as prevailing ocean currents or continental effects — demonstrates contextual understanding and can earn higher marks.

偏态也应当通过比较来阐述。如果某一分布的均值远大于其中位数，而另一分布大致对称，你可以推断前者呈正偏态。提及真实的天气原因——例如主导洋流或大陆性影响——能够体现对背景的理解，并可能带来更高的分数。

10. Handling Missing Values and Outliers in Calculations | 计算中对缺失值和异常值的处理

When a variable contains ‘n/a’, you must exclude that case from calculations of mean, median or standard deviation. The sample size n then refers only to the number of available observations. If a rainfall reading is recorded as ‘tr’, it is usually treated as zero unless the question specifies otherwise. Always state your assumptions clearly.

当某一变量包含 ‘n/a’ 时，你必须将该条记录从均值、中位数或标准差的计算中剔除。此时样本容量 n 仅指可用的观测值个数。如果降雨量的读数被记录为 ‘tr’，除非题目另有说明，通常将其视为零。请务必清楚地说明你的假设。

Outliers can be identified using the rule Q₁ − 1.5 × IQR for lower bounds and Q₃ + 1.5 × IQR for upper bounds. In the LDS, a day with an extremely high maximum gust might be flagged as an outlier. Your task may be to decide whether to include it: if the purpose is to summarise a ‘typical’ month, it might be removed; if the goal is to capture the full range of weather extremes, it should stay.

异常值可以使用下界 Q₁ − 1.5 × IQR 和上界 Q₃ + 1.5 × IQR 的规则来识别。在大型数据集中，某天出现极高的最大阵风数值可能会被标记为异常值。你的任务可能是决定是否将其纳入：如果目的是概括一个“典型”月份的情况，或许可以将其移除；如果目标是捕捉完整的极端天气范围，则应该保留。

11. Exam‑style Questions and Strategies | 考试题型与应对策略

Typical exam tasks include: selecting and justifying a sampling method, cleaning a small extract of the LDS, calculating means and standard deviations from a list or a frequency table, drawing and interpreting box plots or histograms, and writing comparative paragraphs. Some questions give you partially completed summary tables and ask you to fill in missing values using formulae.

典型的考试任务包括：选择并论证一种抽样方法、清理大型数据集的一个小摘录、根据列表或频数表计算均值和标准差、绘制并解读箱线图或直方图，以及撰写比较段落。有些题目会给出部分完成的汇总表格，要求你利用公式填补缺失值。

Read the question data carefully: the LDS extract will include units and sometimes anomalous symbols. Before diving into calculations, scan for ‘n/a’, ‘tr’ and obviously impossible numbers. Write down the values you plan to use for n, Σx and Σx², showing clear working. When asked to compare, use the structure ‘measure × location = value, interpretation’. This systematic approach saves time and reduces errors.

仔细阅读题目中的数据：大型数据集的摘录会包含单位，有时还会有异常符号。在投入计算之前，先扫描一遍 ‘n/a’、’tr’ 以及明显不可能出现的数字。写出你计划用来计算的 n、Σx 和 Σx² 的值，并展示清晰的过程。当被要求进行比较时，使用“度量值 × 地点 = 数值，解读”的结构。这种系统性的方法可以节省时间并减少错误。

12. Summary and Key Points | 总结与要点回顾

The Large Data Set is not a separate topic but a tool that weaves through the entire statistics content of Edexcel A‑Level Mathematics. Familiarity with its variables, units and quirks allows you to approach questions with confidence. Keep a mental checklist: identify data types, clean carefully, choose appropriate measures of location and spread, draw clear diagrams, and always answer in the context of weather.

大型数据集不是一个独立的主题，而是贯穿 Edexcel A‑Level 数学整个统计部分的一项工具。熟悉它的变量、单位和各种特点能让你充满信心地应对问题。在心中牢记一份核对清单：识别数据类型、仔细清洗数据、选择合适的集中趋势和离散程度度量、绘制清晰的图表，并始终在天气背景下作答。

Practice by taking small samples from the official spreadsheet and exhaustively describing them. The more you engage with real data, the more fluent you become in the language of statistics — an ability that will serve you well beyond the examination hall.

不妨从官方电子表格中抽取小样本来进行练习，并详尽地描述它们。你接触真实数据越多，你的统计学语言就会越流利——这种能力将使你在考场之外也受益匪浅。

Published by TutorHao | Mathematics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）

The Large Data Set | 大型数据集