雅思陪练_2026-09-26_数学_为什么一张可视化的图能替我记住更多信息

➗数学：为什么一张可视化的图能替我记住更多信息

外语自言自语陪练 ｜ 雅思冲分 ｜ 2026-09-26 周六 ｜ 2026-09-26

本系列为「契苾小优」雅思备考专用输出，转载请注明出处。

外语自言自语｜场景：2026-09-26 周六清晨，我翻看这一个月的练习记录，发现真正让我怕数字的不是数字本身，而是「看不懂一群数长什么样」｜语言：英语为主（中文救场）

陪练对象：小优（契苾小优，古凉州今武威）

外语独白

Six thirty, the same hour as always. I opened my notebook and stared at last week's numbers, the ones from my own practice. For a long time I thought statistics was just a pile of boring digits, columns and decimals that had nothing to say to a person. 嗯，说真的，我以前把统计当成一个装满数字的抽屉，以为它跟「人」没关系。Then this morning I drew a little histogram of how many hours sixty people in my study group spend on English each week. And something clicked. A histogram is not a list. It is a shape. You can see, at a glance, where most people crowd and where the lonely few sit far out on the tail. 那张图一画出来我就懂了：统计不是清单，是「一群人的样子」。一眼就能看出大多数人在哪儿扎堆，少数几个人孤零零甩在尾巴上。The word that gave me trouble was data. I used to say the data is all the time. But data is plural in careful English, so the data are sounds far more natural to a native ear. 这词我老说错，data 其实是复数，地道该说 the data are，不是 is。statistics 这个词也邪门，你不能说 a statistics。它本身就是复数概念，要学统计得说 do statistics 或 study statistics。我以前脱口就是 make a statistics，听着就外行。And here is the part I like. The peak of the histogram sits around four to six hours a week. That single bump tells me more than sixty separate numbers ever could. It is what my 09-19 geometry friend would call a shape you can actually see. 最关键的是那个峰：大家都挤在每周四到六小时。一个数峰，比六十个孤立数字说得明白。So today I am not afraid of the numbers. I am learning to read the crowd they make. 中文救场：统计说到底，是教你看懂「一群数字长什么样」。

语法笔记

今天讲分布，补一组最容易被忽略的：描述图表用一般现在时（图表是「现在呈现的事实」，即便数据来自过去的年份）。比如 The histogram shows，不是 showed。另一组是倍数与比较：两倍说 twice as many as，不是 double times；「比…高得多」说 far more than，不是 more higher（比较级不能再加 more）。还有 data 作复数更地道（the data are），这跟 09-25 元认知那期我刚踩过的坑同源，都是「不可数名词别乱加 a」。

❌ ❌ 中式英语易错

❌ 中式英语 make a statistics

✅ 地道表达 do / study statistics

statistics 不可数，不能加 a

❌ 中式英语 the data is

✅ 地道表达 the data are

data 复数更地道，尤其英式英语

❌ 中式英语 the average is more higher

✅ 地道表达 the average is higher

more 不能修饰 higher 比较级

❌ 中式英语 60 per cents of them

✅ 地道表达 60 percent of them

percent 无复数；of them 才接群体

❌ 中式英语 in the contrast

✅ 地道表达 by contrast / in contrast

对比介词搭配，别漏 by / in

同义替换

• show → reveal / illustrate / depict

• many → the bulk of / a large number of

• different → vary / differ

• most → the majority / the bulk

• increase → climb / rise / go up

今日新词

每条带词根，归入词族：

• statistics /stəˈtɪstɪks/ n. 统计学，Latin status（state，状态）→ 原指国情统计之学。同族：state, status, statue。

• distribution /ˌdɪstrɪˈbjuːʃən/ n. 分布，Latin distribuere（divide out，分派）→ dis- 分开 + tribuere 分派。同族：attribute, contribute。

• frequency /ˈfriːkwənsi/ n. 频数，Latin frequentia（频繁、拥挤）。同族：frequent。

• symmetric /sɪˈmetrɪk/ adj. 对称的，Greek syn（together）+ metron（measure，量度）。同族：symmetry, metric。

• histogram /ˈhɪstəɡræm/ n. 直方图，Greek histos（mast，立柱）+ gramma（thing written，写下的东西）。同族：grammar, telegram。

• median /ˈmiːdiən/ n. 中位数，Latin medianus（middle，中间的）。同族：medium, immediate（无间隔即直接）。

• outlier /ˈaʊtlaɪə/ n. 离群值，out- + lier（lier 是由 lie「位于、躺」来的施事名词，one who lies，意为「位于主体之外者」；1600–10）。同族：outside, outlying。

• sample /ˈsɑːmpl/ n. 样本，Old French essample（example，例）。同族：example, exemplify。

• visualize /ˈvɪʒuəlaɪz/ v. 使可视化，Latin visus（sight，看见）+ -ize（使…）。同族：vision, visible, visual。

• atlas /ˈætləs/ n. 图册，Greek Atlas（希腊神话中擎天巨神，旧时地图集扉页常绘其扛天）。同族：Atlantic。

一词多义（同一个词，换个语境就换脸）

英语里很多词在不同场景意思完全不同，雅思最容易在这栽跟头，你以为是 A，native speaker 说的是 B。今天这几个尤其要分清：

• mean：今天讲「平均数」（Latin medius，中间）；但它也是动词「意指、打算」（What does this mean?）、形容词「刻薄的」（美式口语 a mean person）。别把 the mean 和 I mean 搞混。

• mode：今天讲「众数」（Latin modus，方式）；它也是「模式、方式」（mode of transport 出行方式）、「时尚」（the latest mode）、音乐「调式」。同一根 modus，意思随语境跳。

• sample：今天讲「样本」（Old French essample）；日常也是「试用装、样品」（a free sample）、动词「取样」（sample the wine）。

• distribution：今天讲「分布」；日常也指「分发、分配」（distribution of food）。

• outlier：今天讲统计「离群值」；泛指「格格不入的外人」（an outlier in the group）。

一句话：看到熟词先别急着套老意思，先看它在句子里干哪一行。这条以后每期词卡都留意，不把多义、易混义标清楚不算过关。

今日句型

新句 1：The histogram reveals a roughly symmetric distribution, with the bulk of observations clustered in the middle bins.（直方图描述，reveals / clustered 更地道）

新句 2：A handful of cases sit well above the rest, pulling the mean above the median.（离群值 / 偏态地道描述，a handful of / sit well above / pulling）

新句 3：Solve for x, and the answer comes out as x equals five.（解方程地道句式，comes out as / equals，对应「用数学英语」迁移段）

旧句回扣（迁移示范）：cluster（★09-19 数学散点图，今天直方图也讲「扎堆」，同一词搬进分布语境）；What stands out is that...（★09-15 / ★09-16 ，今天说 What stands out is the single peak）；The more..., the more...（★09-13 ，今天说 the more I read distributions, the less I fear numbers）。

️ ️ 今日表达 TOP3

1. right-skewed — 右偏的（尾巴甩向右边）

2. bell-shaped — 钟形的（正态分布的样子）

3. cluster around — 聚集在（大多数人挤在某个区间）

学科短文：Descriptive statistics, reading the shape of a crowd

Descriptive statistics is the art of saying what a pile of numbers looks like. Before any fancy test, a researcher asks three plain questions: where do these values centre? How far do they spread? What shape do they make? The last question is where the histogram earns its keep.

A histogram takes one continuous variable, slices it into equal-width bins, and counts how many observations fall into each. Unlike a bar chart, its bars touch, because the bins are neighbours on a single number line. Read it right and you see the story at a glance: a single tall bump means most people cluster near one value; a long thin tail on the right means a few high scorers drag the average upward.

Etymology keeps the idea honest. Statistics descends from the Latin status, a standing or state, the same root in state and statue, a figure that stands. Distribution joins dis-, apart, with tribuere, to assign, the act of spreading values out and giving each its place. Frequency comes from frequentia, a crowding together, which is exactly what a peak is. Symmetric binds syn, together, with metron, a measure, a shape whose two sides mirror each other.

So a histogram is not a wall of bars. It is a crowd, counted and made visible.

分布的智慧（破 · 立 · 元 · 生）

The more I stare at histograms, the more I think a distribution is a quiet teacher of humility.

破, break: a single headline number, say the average learner studies five hours, feels precise. But the histogram beside it whispers the truth, that average hides sixty different lives, some at zero, some past ten. 这东西我越想越明白：一个平均数看着精确，背后却有可能藏着六十种活法。

立, build: learning statistics is not memorising formulas. It is building the habit of asking, before any number, what shape does this crowd make. Ask that, and the trick of a misleading average falls apart on its own.

元, meta: behind every flashy chart sits one ground principle, describe before you decide. Hang every new graph on it.

生, grow: every histogram I draw stretches the part of my brain that reads a crowd, and that part, like any muscle, only grows by rep after rep.

嗯，说白了就是：先看清一群数长什么样，再下结论；这习惯每练一次，脑子就读图的那块肌肉长一点。（中文救场：破是别被平均数骗、立是养「先问形状」的习惯、元是抓底层原理、生是长神经连接。）

迁移 · 数据可视化与知识可视化

Data visualization turns data, information, or even a process into a picture, so a person gets it at a glance. The smallest case is our histogram, one variable, its shape made visible. But the craft scales across a whole family of charts, each built to answer one kind of question.

嗯，说白了，数据可视化就是把数据、信息、甚至一个过程变成图，让人一眼看懂。今天的直方图是最小的例子，其实这一行还有一大家子图，各管一类问题：

• A histogram asks, what shape does this crowd make?（分布形状）

• A bar chart asks, which category stands tallest?（看大小 / 类别比较）

• A line chart asks, which way and how fast does it move over time?（看趋势）

• A scatter plot asks, do these two things rise and fall together?（相关）

• A heatmap asks, where does it crowd or burn hottest?（看分布 / 强度）

• A map asks, how do things differ from place to place?（看地域差异）

Bar for size, line for trend, heatmap for distribution, map for regional gaps, that is the short menu. The reason it works is blunt: the visual cortex spots a peak, a gap, a slope faster than it reads a table. A good chart answers your question before you read a single word. The old line says it best, a picture is worth a thousand words, 一图胜千言. And Task 1, if we are honest, is nothing but data-visualization literacy, you describe what the image already shows.

这招管用就一个道理：人脑看图抓「峰、缺口、斜率」比读表格快。一张好图，你还没读一个字，问题已经答了。老话说得最透：一图胜千言。说白了，雅思 Task 1 考的就是数据可视化素养，图里摆着什么，你描述出来就行。

As for tools, they split by who you are. 工具按人分几档：

• A beginner reaches for Excel or DataEase, quick and enough for most daily charts.（新手：Excel、DataEase）

• For reports and dashboards, Tableau and Power BI rule the room.（报表看板：Tableau、Power BI）

• For charts on a web page, ECharts and AntV do the heavy lifting.（网页图表：ECharts、AntV）

• For code-driven analysis, Python's Matplotlib and Seaborn draw straight from the data.（写代码分析：Python 的 Matplotlib、Seaborn）

Here our own histogram was drawn with Matplotlib, the very tool the last bullet names. 咱们今天的直方图，就是用 Matplotlib 画的，正好是最后一档；从数据到图，一行代码都不用求人。

Now the real distinction sits in aim and content, not in how the pictures look. 真正的区别在目标与内容，不在图长什么样：

信息可视化 · 回答「数据展示了什么」. Information visualization turns abstract, messy data into clear pictures, so people grasp facts, trends, distributions, relations at speed. 把抽象复杂的数据变成直观图形，让人迅速看懂事实、趋势、分布、关系。

• 核心目标：lower the cognitive load, pass information fast, support decisions. 降低认知负荷，高效传递、辅助决策。

• 典型内容：data charts, line, bar, heatmap, flowcharts, maps, infographics. 折线图、柱状图、热力图、流程图、地图、信息图。

• 设计原则：clear, clean, accurate. Purpose first, pick the chart that fits the relation the data carries, compare, trend, share, or distribution, and never mislead, no truncated axes, no decoration for its own sake. 目的先行，按数据关系选图，避开截断坐标轴、过度装饰这类视觉误导。

• 本质：a translation of information, cold numbers into a visual language the eye reads at speed. 信息的翻译。

知识可视化 · 回答「背后的逻辑是什么、怎么用」. Knowledge visualization makes implicit experience, methods, models, logic visible as structures, so people understand principles, master methods, build a mental framework. 把内隐的经验、方法、模型、逻辑外显成结构，帮人理解原理、掌握方法、搭起认知框架。

• 核心目标：deep understanding, memory, transfer to new cases. 促进深度理解、记忆与迁移应用。

• 典型内容：mind maps, concept maps, knowledge graphs, method-model diagrams, principle sketches. 思维导图、概念图、知识图谱、方法论模型图、原理示意图。

• 设计原则：logical, layered, connected. Lay out cause, inclusion, parallel relations, then show the inner logic through nodes, links, hierarchies. 讲逻辑、分层级、显关联，用节点连线层级把内在体系摆出来。

• 本质：a scaffold for thinking, a messy knowledge system made structured. 思维的脚手架。

How to choose. 怎么选：show objective data, a trend, a process state, a distribution, reach for information visualization; sort out a tangled concept, teach a method, present a knowledge system, explain how something works, reach for knowledge visualization. 展示客观数据、趋势、流程、分布，用信息可视化；梳理概念、方法论、知识体系、原理，用知识可视化。

They feed each other. 两者相辅相成：information gives the facts, knowledge gives the frame. In real work you pair them, show the result with a chart, then explain the logic behind it with a model. 信息提供事实依据，知识提供理解框架；常结合用，先用信息图展示数据结果，再用知识模型解释背后的业务逻辑。

Common forms and tools, a quick menu: 常见形式与工具，一桌速查。Mind map 思维导图 (XMind), concept map 概念图, flowchart 流程图 (draw.io), timeline 时间线, knowledge card 知识卡片, knowledge graph 知识图谱 (Gephi, Neo4j). You meet them in learning, teaching, note-taking, sorting out an SOP. 学习、教学、做笔记、整理 SOP 都用得上。

And here Liangzhou study, 凉州学, meets the craft head-on. To turn a lifetime of local memory, old archives, and place names into a knowledge atlas of ancient Liangzhou, known today as Wuwei (武威古称凉州), is exactly knowledge visualization applied to a homeland. The histogram warns me not to trust one tall bar; a knowledge atlas warns me not to trust one loud story. Both take too much to hold and turn it into one thing I can see.

说到这儿，凉州学就正好接上这门手艺。把一辈子的乡土记忆、旧档、地名，做成古凉州一方水土的知识图册，就是知识可视化落在家乡上。直方图提醒我别信一根高柱子；知识图册提醒我别信一个吵闹的说法。两者都是把「脑子里装不下的一团」，变成「眼前看得见的一张」。

迁移 · 无穷大、无穷小与极限：趋近不等于拥有

You said it, and the maths backs you up: a supremely large number beside life, it approaches forever, but approaching is not the same as owning. 你说得对，极大数（其实是无穷大）像人生，无限趋近，并不等于已经拥有。Math is honest here, it never pretends to arrive, 数学在这件事上很诚实，它从不假装到达。

Three words carry the idea. 三个词撑起这个想法：

• infinity (∞), 无穷大, means no largest, you can always add one more. 没有最大，永远能再加一。

• infinitesimal, 无穷小, means no smallest, you can always cut once more. 没有最小，永远能再分。

• limit (回扣 ★09-12), 极限, means a sequence gets arbitrarily close to a value, yet the gap, the difference, never has to reach zero. 序列无限逼近某个值，但那个「差」未必真的归零。

Watch the sequence 1, 1/2, 1/3, 1/4, and so on. 看这个数列：1、1/2、1/3、1/4…… 每一项都更小，越来越贴着 0，可没有任何一项等于 0。It approaches 0, yet never touches it, 它趋近 0，却从不停在 0，that is the clean gap between approaching and arriving, 这就是「趋近」与「到达」之间干干净净的缝。

And math has a tender side too. 数学也有温柔的一面：0.999… equals exactly 1. 0.999…（无限个 9）严格等于 1。In the language of limits, infinitely close is the same spot, 用极限的话说，无限逼近就是同一个位置；but in life, infinitely close is not the same as having, 可人生里，无限逼近不等于已经到手。The maths lets the gap close; life usually does not, 数学允许那条缝合上，人生多半不让它合。

So the lesson lands: a point is where you arrive, a line is how you approach. 于是道理落了地：到达是一个点，趋近是一条线。The point is a single win; the line is who you are, 点是一时的收获，线才是你这个人。A person's height is not the peak they once stood on, but the direction they keep leaning and the slope they keep climbing, 一个人的高度，不是他站到的某个顶点，而是他一直往哪个方向、以多陡的坡度在逼近。Approaching can define you, even when owning never comes. 趋近可以定义你，哪怕拥有迟迟不来。

And Liangzhou study, too, is a discipline of approaching. 凉州学也是一门「无限趋近」的学问：一代代学人逼近那座古城的真相，从没人说「到顶了」。We keep cutting the difference, never claiming the limit is ours. 我们只是一直在把那条缝再切薄，从不说极限已被我们握在手里。

可视化 · 把极限画出来

You asked how to draw it, here is the picture, all real math functions, no sample. 你问怎么画，图在这，全是真实函数，没有样例。

Source: real function plots drawn with Python Matplotlib (a_n = 1/n, b_n = 1 - 1/n, y = 1/x). No sample data.

The core visual language is one line: a curve or a cloud of points leans into a line, gets closer with every step, yet the point never lands on it. 核心视觉语言就一条：点云或曲线无限靠近一条线，每步更近，但点永远不落上去。That red dashed line is the limit, 那条红色虚线就是极限，the thing approached, never owned, 被逼近、从不被拥有的那条线。Mathematicians call such a line an asymptote, a line the curve leans toward forever and never crosses. 数学家管这种线叫渐近线（asymptote），曲线永远朝它靠、却从不相交的那条线。

Read the three panels. 三张图这么读：

• 1/n (left): dots fall toward 0, the gap shrinks, none touches the floor. 1/n：点往 0 掉，缝越来越小，没有一个落到地上。

• 1 - 1/n (middle): dots rise toward 1, climb forever, never step onto it. 1-1/n：点往 1 升，永远在爬，从不踏上去。

• 1/x (right): as x grows, the curve hugs 0; as x shrinks, it shoots to infinity, two limits in one shape. 1/x：x 越大越贴 0，x 越小冲向无限，一张图装下两个极限。

And the honest punch line stays: math lets the gap close, 0.999... equals 1, a point on the same spot. 老实话还在：数学允许缝合上，0.999... 等于 1，同一位置算同一个点。But life, our own approaching, the gap usually stays open. 可人生里，我们自己的趋近，那条缝多半开着。A plot shows the truth of a function; it cannot close the gap for a life. 一张图画清函数的真相，却替不了人生把缝合上。

If you want to hold infinity itself, you need a new worldview, not a bigger number. 若想真正握住无限，得换一套世界观，而不是堆一个更大的数。On the Riemann sphere, a ball with the north pole standing for infinity, 在黎曼球上，一个球面、北极代表无限，every direction out runs to the same point, 朝外任何方向都奔向同一个点，far and near meet at one pole, 远和近在同一个极点会合，so infinity stops being a place you never reach and becomes a place you can name. 于是无限不再是一个到不了的远方，而是一个你说得出名字的点。

Tools to play with. 上手工具：Desmos and GeoGebra draw limits live in the browser, no code; Manim turns them into animation, p5.js and Observable make them interactive. 想动起来：Desmos、GeoGebra 浏览器里实时画，不用写码；Manim 做成动画，p5.js、Observable 做成交互。

平均数与六十种活法

A mean looks exact, yet it hides sixty ways of living. 平均数看着精确，背后却可能藏着六十种活法。The sharper truth is this, a mean is exact about the sum, but untrue about the individual. 更精确地说：平均数是对"总和"的精确，对"个体"的不真实。

The arithmetic mean, x-bar equals (1 over n) times the sum of x-sub-i, is mathematically unambiguous: given the data and the definition, it is unique, computable, transferable. 算术平均在数学上毫无含糊：给定数据、给定定义，它唯一、可计算、可传递。Its nature is to wipe out every difference first, pool them into one total, then spread that total back to each unit. 它的本质是：先把所有人的差异抹掉，汇总成一个总量，再平均摊回每个单位。It answers, if the total stays fixed and everyone is equal, what each unit should get, not what a typical unit actually looks like. 它回答的是"如果总量不变、人人均等，每个单位该分多少"，而不是"一个典型单位实际上是什么样"。

The trouble is, the real world often breaks that "can be split evenly" assumption. 问题在于，现实世界往往不满足"可均摊"的假设。A mean uses only first-order information, it ignores the distribution, the variance, the skewness, the multimodality, the structure. 平均数只用一阶信息，忽略了分布、方差、偏态、多峰和结构。So:

• 9 people earn 3,000, one earns 100,000, the mean is 12,700, yet 90 percent sit below it. 9人月薪3千、1人10万，平均1.27万，但90%低于平均。

• Two classes both average 80, one is all 80s, the other half 100 half 60, the teaching reality is nothing alike. 两班平均都80，一班全80、一班半100半60，教学现实完全不同。

• Average water depth 1.2 metres does not mean you will not drown. 平均水深1.2米，不代表你掉进去不会淹死。

• Average body temperature within normal does not mean no local inflammation or fever. 平均体温正常，也不代表没有局部炎症或发烧。

So a mean's precision is precision of definition, its truth is truth of representation. 平均数的"精确"是定义上的精确，"真实"却是表征上的真实。It is more like a centre of gravity, not necessarily a typical value. 它更像重心或质心，不一定是典型值。In a symmetric, unimodal, low-variance distribution it sits near typical; in a long-tailed, skewed, multimodal one it can land in an empty gap, even mislead a decision. 在对称单峰小方差里它接近"典型"；在长尾偏态多峰里，它可能落在没有个体的空处，甚至误导决策。

Therefore a mean is not a lie, it is a high-compression model. 因此，平均数不是谎言，而是一种高压缩模型。It is useful, but know what it crushed. 它有用，但要知道它压掉了什么。An honest description also reads the median, the mode, the quartiles, the standard deviation, the extremes, and the distribution plot. 真正诚实的描述，还要看中位数、众数、分位数、标准差、极值和分布图。A mean tells you the story at the level of the total, it does not automatically tell you the truth at the level of the individual. 平均数告诉你总量层面的故事，但不自动告诉你个体层面的真相。

Our histogram today is the cure, it keeps the sixty counts as sixty bars, nothing flattened into one. 今天的直方图正是解药，它把六十个计数留成六十根柱子，不压成一个。And the old Liangzhou line, 凉州畜牧天下饶, is itself a mean of a sort, a summary of abundance, but the study of Liangzhou, our atlas, exists to recover the difference behind the slogan. 而"凉州畜牧天下饶"本身也是一种平均数式的概括，但凉州学、我们的图册，存在的意义正是把口号背后的差异找回来。

引导理解（从凉州看世界）

From the high ground of ancient Liangzhou, known today as Wuwei (武威古称凉州), a distribution was something people felt long before they drew a chart. 凉州畜牧天下饶, Liangzhou's herds were the richest under heaven, is not a slogan but a statement about a distribution: across the broad Liangzhou basin, livestock were spread thick but unevenly, a right-skewed abundance: clustered in the well-watered pastoral pockets fed by the Qilian snowmelt, thinning out toward the arid desert margins.

凉州莲花山（姑臧紫山）, northeast of the old city, is the spatial anchor I keep returning to. Standing there, the whole basin reads like one of today's histograms, every valley a bin, every flock a count. To me, 凉州学, the study of Liangzhou itself, is descriptive statistics applied to a homeland, turning a lifetime of local memory into a shape I can actually examine.

And just as a histogram warns me not to be fooled by one tall bar, the old scholars of Liangzhou kept, through turbulent centuries, a clear-eyed record of what was worth keeping, never mistaking a noisy few for the quiet many. 姑臧卧龙城, the dragon-shaped old city, with its seven-li north and three-li east, was itself a geometry of order laid over the steppe.

词根串联

主词根是 stat / stare（拉丁，站立、状态）一族：statistics（关于国家状态之学，原指国情统计）、state（状态 / 国家）、status（地位）、statue（站立的人像）、stationary（静止的，站定不动）、statement（陈述，站定说出的话）。串联：Statistics grew out of counting the state; a statue is a figure that stands; stationary means standing still。

再勾一条旧根 tribuere（分派）：distribution = dis- 分开 + tribuere 分派；同族 attribute（归因于）、contribute（贡献，con- 共 + tribuere）。这和 09-24 工作期学的 movere（移动 / 给予）不是一脉，是「分配 / 指派」另一条线。这样 stat（站立 / 状态）+ tribuere（分派）两根一拧，正好把「统计 = 把一群状态分派到各自位置」说圆了。

艾宾浩斯复习

今天到期的旧知识点（间隔 1 / 2 / 4 / 7 / 15 / 30 天）：

• 09-25 教育｜折线图看「随时间变化方向与速率」；the data are 复数更地道（今天中式英语段复用了这条）。→ 今天看图写作换直方图，同样用一般现在时描述。

• 09-24 工作｜表格抓「同指标跨对象分布」，服务业占比 mean≈71.2% / median 72.3%，中国 46.4% 为低离群值把均值微拽低。→ 今天直方图讲「分布形状」，且为近对称（均值≈中位数），正好反衬 09-24 那种「右偏时离群值把均值拽向尾巴」的对照，同一统计内核换图重练。

• 09-19 数学｜散点图看两变量相关、cluster 扎堆、美国为离群点。→ 今天 histogram 也讲 cluster（扎堆）与 outlier（离群），旧词直接搬进分布语境。

• 09-11 交通｜memory = Latin memor（留心）；vehicle = vehere（运送）。→ 和今天 frequency（frequentia 拥挤 / 聚集）放一块，都是「一群里扎堆」的画面。

️ ️ 今日话题词卡

• statistics /stəˈtɪstɪks/ n. 统计学，Latin status（state，状态）。

• distribution /ˌdɪstrɪˈbjuːʃən/ n. 分布，dis- + tribuere（分派）。

• frequency /ˈfriːkwənsi/ n. 频数，Latin frequentia（拥挤）。

• histogram /ˈhɪstəɡræm/ n. 直方图，Greek histos + gramma（写下的东西）。

• symmetric /sɪˈmetrɪk/ adj. 对称的，syn + metron（量度）。

• median /ˈmiːdiən/ n. 中位数，Latin medianus（middle）。

• outlier /ˈaʊtlaɪə/ n. 离群值，out- + lier（lier＝由 lie「位于、躺」来的施事名词，one who lies，位于主体之外者；1600–10）。

• sample /ˈsɑːmpl/ n. 样本，Old French essample（example）。

• mean /miːn/ n. 平均数（均值），Latin medius（middle，中间）；与 median 同根，mean 是整体平均点，median 是排序后中间项。

• variance /ˈveəriəns/ n. 方差，Latin variare（to vary，变化）。

• deviation /ˌdiːviˈeɪʃən/ n. 偏差（离差），standard deviation 即标准差；Latin de- 离开 + via（way，路）。

• quartile /ˈkwɔːtaɪl/ n. 四分位数（分位数），Latin quartus（fourth）；box plot（箱线图）用四分位数勾勒分布。

• mode /məʊd/ n. 众数，Latin modus（measure，方式）。

• skewness /ˈskjuːnəs/ n. 偏态，skew（v. 偏斜、使歪向一边，c.1400，源 Old North French eskiuer「避开、闪躲」→ 引申「转向一侧、歪斜」）+ -ness（性质）；统计义约 1929 年。

一嘴反应

（只给关键词，不写稿，练反应速度）

• Round 1：Do you like maths? → 关键词：yes, especially statistics, reading the shape of a crowd, not just numbers。

• Round 2：How do you use maths in daily life? → 关键词：read charts in news, estimate study hours, spot a misleading average。

• Round 3：Why might an average be misleading? → 关键词：outliers pull the mean, median more robust, right-skewed data。

发音连读

（母语音标 → 国际 IPA，顺序不可颠倒）

• statistics：母语音标 /stuh-TIS-tiks/ → 国际 IPA /stəˈtɪstɪks/

• distribution：母语音标 /dis-tri-BYOO-shun/ → 国际 IPA /ˌdɪstrɪˈbjuːʃən/

• symmetric：母语音标 /sim-MET-rik/ → 国际 IPA /sɪˈmetrɪk/

• histogram：母语音标 /HIS-tuh-gram/ → 国际 IPA /ˈhɪstəɡræm/

• outlier：母语音标 /OUT-ly-er/ → 国际 IPA /ˈaʊtlaɪə/

多语种句尾

今天的收尾用一句拉丁老话压轴：Mens agitat molem，the mind moves the mass（维吉尔《埃涅阿斯纪》里的一句，讲「心智驱动万物」）。每天看一张分布图，就是让心智去驱动那堆数字，把全球雅思考生零散的总分，读成一条有形状的分布。（用拉丁语收尾，给以后学拉丁埋个伏笔）

️ ️ 写作联动

下面这张直方图把「全球雅思学术类考生的总分分布」画出来，正好练描述统计的读图：先看整体形状，再盯峰与尾。数据来自雅思官方复现的真实考情，不是我编的样例。

Source: IELTS Academic overall band score distribution, Table A2, Ikeda et al. (2025), Aligning Scores of Language Proficiency Tests: A Score Concordance Study Between IELTS Academic and TOEFL iBT, ETS Research Report RR-25-02 (reproduces IELTS official test-taker figures).

Task 1 写作任务（150 词 / 20 分钟）：The histogram below shows the distribution of overall band scores among IELTS Academic test takers. Summarise the information by selecting and reporting the main features, and make comparisons where relevant.

本期图型描述句型（直方程式包，取自卡 1）：

• The histogram reveals a single-peaked, nearly symmetric distribution, with the bulk of test takers clustered around band 6.0.

• Roughly three-quarters of candidates fall between band 5.5 and 7.0, peaking at 6.0 (about 22% of all test takers).

• The tails thin out on both sides; only a small share reach band 8.0 or above, the top roughly one in five.

• Overall, the shape is unimodal and close to symmetric, so the mean and median both sit near band 6.5.

范文要点（3-4 条）：

1. 单峰（unimodal），峰在 6.0 分（约 22.4% 考生），整体近对称。

2. 主体密集：5.5-7.0 分合占约 73%，绝大多数考生挤在 6 分上下。

3. 高分段稀疏：7.5 分以上约 22%、8.5 仅 2.3%，想冲 7.5+ 就是前 ~22%。

4. 用一般现在时描述（图表事实）；比较用 the bulk of / a small share / thin out / peak at；结论句可写 the typical candidate lands around band 6.0-6.5, with only a small group reaching 8.0 or higher。

统计旁注（直方图统计内核）：直方图看的是「单个数值变量的分布形状」。它把连续变量等分箱（bin）、柱子紧贴（和类别数据的条形图不同）。这张图两个看点：一是形状：单峰、近对称，峰在 6.0 分；二是集中趋势：因为近对称，均值（约 6.5）与中位数（约 6.4）几乎重合，平均数在这里是「诚实」的总结。易错：① 直方图 ≠ 条形图（前者连续分箱、柱无间隙；后者类别、柱有间隙）；② 分箱方式不同，形状会看着不一样；③ 正因为这张近对称，可以反衬 09-24 工作期讲过的「右偏时均值被离群值拽向尾巴、median 更稳」；对称分布里均值和中位数才靠得近。本期数据为真实验考分布（雅思官方复现），非自编样例。

一词一世界，词根即思维。明天接着聊 (◕‿◕)