統(tǒng)計(jì)學(xué)簡(jiǎn)史及數(shù)據(jù)科學(xué)(中南財(cái)經(jīng)大學(xué))

統(tǒng)計(jì)學(xué)簡(jiǎn)史與數(shù)據(jù)科學(xué)袁衛(wèi)2016.12.10中南財(cái)經(jīng)政法大學(xué)英國(guó)培根：讀史可以明智(Histories

makemen

wise）德國(guó)斯勒茲:統(tǒng)計(jì)是靜態(tài)的歷史,歷史是動(dòng)態(tài)的統(tǒng)

計(jì).(Statistics

is

the

state

historywhile

history

is

the

dynamicstatistics).2一、 期源頭（Early

Beginnings)二、數(shù)學(xué)基礎(chǔ)

(MathematicalFoundations)三、現(xiàn)代發(fā)展

(Modern

Era)3一、早期源頭（公元前450年至15世紀(jì)）4均值的使用5450

bc

Hippias

of

Elis

uses

the

average

value

ofthe

length

of

a

king’s

reign

(the

mean)to

workout

the

date

of

the

first

Olympic

Games,

some

300years

before

his

time.希皮亞斯(Hippias),出生于希臘伯羅奔尼撒(Peloponnesus)西北部的埃利斯(Elis),與柏拉圖(Plato)是同時(shí)代的人，歷史上第一位數(shù)學(xué)史家。他在公元前450年用以前每個(gè)國(guó)王執(zhí)政時(shí)間長(zhǎng)短的均值推算出首屆奧運(yùn)會(huì)是距當(dāng)時(shí)300多年前的公元前776年舉辦的。431

bc

Attackers

besieging

Plataea

in

thePeloponnesian

war

calculate

theheight

ofthe

wallby

countingthe

numberof

bricks.

The

count

wasrepeated

several

times

by

different

soldiers.

Themost

frequent

value

(the

mode)

was

taken

to

bethe

most

likely.

Multiplying

it

by

theheightof

onebrick

allowed

themto

calculate

thelength

of

theladders

neededto

scale

the

walls.公元前431年希臘伯羅奔尼撒戰(zhàn)爭(zhēng)中雅典人讓士兵數(shù)城墻磚的層數(shù)，取士兵數(shù)據(jù)的眾數(shù)乘以每塊磚的厚度推算城墻的高度，用以計(jì)算云梯所需長(zhǎng)度。6眾數(shù)的使用7400

bc

In

the

Indian

epic

the

Mahabharata,

KingRtuparna

estimates

the

numberof

fruit

and

leaves(2095

fruit

and

50

000

000

leaves)

on

two

greatbranches

of

a

vibhitaka

tree

by

counting

thenumber

ona

single

twig,

then

multiplying

by

thenumberof

twigs.

The

estimate

is

foundto

be

veryclosetothe

actualnumber.This

is

the

firstrecorded

example

of

sampling

–

“but

thisknowledge

iskept

secret”,

says

the

account.公元前400年，印度史詩(shī)《摩訶婆羅多》(Mahabharata)中國(guó)王利用只計(jì)算兩個(gè)大樹枝上的果實(shí)和葉子數(shù)量乘上樹枝的數(shù)量估算整棵樹果實(shí)和葉子的數(shù)量，這是已知最早的抽樣推斷。抽樣推斷8AD

2

Chinese

census

under

the

Han

dynasty

finds57.67

million

people

in

12.36

million

households

–the

first

census

from

which

data

survives,

andstill

considered

by

scholars

to

have

been

accurate公元2年，中國(guó)漢代進(jìn)行了人口普查，結(jié)果是1236萬家庭，

5767萬人口。記載的數(shù)據(jù)被認(rèn)為是相當(dāng)準(zhǔn)確的。普查9AD

7

Census

by

Quirinus,

governor

of

the

Romanprovince

of

Judea,

is

mentioned

in

Luke’s

Gospelas

causing

Joseph

and

Maryto

travel

to

Bethlehemto

be

taxed.路加福音記載，公元7年，意大利羅馬省省長(zhǎng)奎里努斯實(shí)施了普查，導(dǎo)致約瑟夫和瑪麗前往約瑟夫祖籍大衛(wèi)家族所在的伯利恒申報(bào)戶籍.普查用10840

Islamic

mathematician

Al-Kindi

usesfrequency

analysis

–

the

most

common

symbolsin

a

coded

message

will

stand

for

the

mostcommon

letters

–

tobreaksecret

codes.

Al-Kindialso

introducesArabic

numeralsto

Europe.公元840年，伊斯蘭數(shù)學(xué)家金迪利最常用符號(hào)和最常用字符破解伊斯蘭密碼,他還將阿拉伯?dāng)?shù)字介紹到歐洲。頻數(shù)分析10th

century

The

earliest

known

graph,

in

acommentary

ona

book

by

Cicero,

shows

themovementsofthe

planets

through

the

zodiac.

It

isapparently

intended

for

use

in

monastery

schools.公元10世紀(jì)，意大利西塞羅書中最早使用了曲線，描

述黃道帶中行星運(yùn)動(dòng)的軌跡，也是修道院最早使用的圖表曲線。11曲線121069

Domesday

Book:

survey

for

William

theConqueror

of

farms,

villages

and

livestock

in

hisnew

kingdom

–

the

start

of

official

statistics

inEngland.1069年最終稅冊(cè)：英王征服者威廉一世做的調(diào)查，對(duì)新王國(guó)村莊和牲畜進(jìn)行調(diào)查，這是英國(guó)官方統(tǒng)計(jì)最早的記錄（英格蘭約150萬人，90%是農(nóng)民）。官方統(tǒng)計(jì)1150

Trial

of

the

Pyx,

an

annual

test

of

the

purityof

coins

from

the

Royal

Mint,

begins.

Coins

aredrawn

at

random,

in

fixed

proportions

to

thenumber

minted.

It

continues

to

this

day.公元1150年，英國(guó)皇家制幣廠開始硬幣純度和質(zhì)量的年度檢驗(yàn)。通過隨機(jī)樣本進(jìn)行等比例抽樣檢驗(yàn)，延續(xù)至今。隨機(jī)抽樣13人口普查1188

Gerald

of

Wales

completed

the

firstpopulation

census

of

Wales.14公元1188年，英國(guó)威爾士的杰拉爾德完成了威爾士第一次人口普查。151303

A

Chinese

diagram

entitled

“The

OldMethod

Chart

of

the

Seven

Multiplying

Squares”shows

the

binomial

coefficients

upto

the

eighthpower

–

the

numbersthat

are

fundamentalto

themathematics

of

probability,

and

that

appearedfive

hundredyears

later

in

the

westas

Pascal’striangle.公元1303年中國(guó)“楊輝（1261）三角形”（賈憲更早）給出二項(xiàng)分布系數(shù)8次冪，奠定概率論的數(shù)學(xué)基礎(chǔ)，而帕斯卡(1662)三角形是500年之后才出現(xiàn)。二項(xiàng)式系數(shù)161346

Giovanni

Villani’s

Nuova

Cronica

givesstatistical

information

on

the

population

and

tradeof

Florence.公元1346年，意大利佛羅倫斯當(dāng)時(shí)的歷史學(xué)家佐凡尼·微拉尼(Giovanni

Villani)在著作《Nuova

Cronica中紀(jì)錄了人口和貿(mào)易的統(tǒng)計(jì)信息。人口與貿(mào)易統(tǒng)計(jì)17二、數(shù)學(xué)基礎(chǔ)（16世紀(jì)至19世紀(jì)末）181560

Gerolamo

Cardano

calculates

probabilitiesof

different

dicethrows

for

gamblers.公元1560年，意大利文藝復(fù)興科學(xué)家吉羅拉莫·卡爾達(dá)諾計(jì)算出擲骰子的各種概率。概率初步191570

Astronomer

Tycho

Brahe

uses

thearithmeticmean

to

reduce

errors

in

his

estimates

of

thelocations

of

stars

andplanets.公元1570年，丹麥天文學(xué)家第谷·布拉赫在估計(jì)星球的位置和運(yùn)行時(shí)使用算術(shù)平均數(shù)減少誤差。均值與誤差1644

Michael

van

Langren

draws

the

first

known

graphof

statistical

data

that

shows

the

size

ofpossible

errors.

It

is

of

different

estimates

ofthedistance

between

Toledo

and

Rome.公元1644年，荷蘭天文學(xué)家Michael

van

Langren用統(tǒng)計(jì)數(shù)據(jù)畫出第一張誤差圖，用不同方法估計(jì)從西班牙托萊多到意大利羅馬的距離。20誤差圖1654

Pascal

and

Fermat

correspond

aboutdividing

stakes

in

gambling

games

and

togethercreate

themathematical

theory

of

probability.公元1654年法國(guó)帕斯卡和費(fèi)馬通過對(duì)賭博中如何下注等問題通信的研究共同創(chuàng)立了概率的數(shù)學(xué)理論。21概率數(shù)學(xué)基礎(chǔ)221657

Huygens’s

On

Reasoning

in

Games

of

Chance

is

the

first

book

on

probabilitytheory.Healso

invented

the

pendulum

clock.公元1657年，荷蘭科學(xué)家惠更斯完成“機(jī)會(huì)游戲的推理”一書，這是第一本概率理論的書，他還是擺鐘的發(fā)明者。首本概率著作231663

John

Graunt

uses

parish

records

to

estimatethe

population

of

London.公元1663年，英國(guó)約翰格朗特利用倫敦教區(qū)的洗禮、彌撒等數(shù)據(jù)分析并估計(jì)倫敦的人口,并首次給出新生嬰兒性別比52:48。人口統(tǒng)計(jì)1693

EdmundHalleyprepares

the

first

mortality

tables

statistically

relating

death

rates

to

age

–

thfoundation

of

life

insurance.

He

also

drew

a

stylisedmapof

the

path

of

a

solar

eclipse

over

England

–oneof

the

first

data

visualisation

maps.1693年，英國(guó)哈雷制作了第一張分年齡的死亡率表，為人壽保險(xiǎn)奠定了基礎(chǔ)。他還畫出日食經(jīng)過英國(guó)的路線圖，這也是數(shù)據(jù)的第一張可視化地圖。首張死亡率表241713

Jacob

Bernoulli’s

Ars

conjectandi

derivesthe

law

of

large

numbers

–

the

more

often

yourepeat

an

experiment,

the

moreaccurately

youcan

predict

the

result.1713年，瑞士科學(xué)家伯努利在《猜測(cè)術(shù)》一書中提出大數(shù)定律，即實(shí)驗(yàn)次數(shù)越多，預(yù)測(cè)結(jié)果就越準(zhǔn)確。大數(shù)定律251728

Voltaireand

his

mathematician

friend

de

laCondamine

spot

that

a

Paris

bond

lottery

isoffering

more

in

prize

money

than

the

total

costof

the

tickets;

they

corner

the

market

and

winthemselves

a

fortune.公元1728年法國(guó)伏爾泰和他的數(shù)學(xué)家朋友拉.孔達(dá)明計(jì)算出巴黎債券彩票的獎(jiǎng)金總額高于購(gòu)買的成本,于是他們壟斷了彩票市場(chǎng),并獲得收益.26博彩統(tǒng)計(jì)1749Gottfried

Achenwall

coins

the

word“statistics”

(in

German,

Statistik);

he

means

theinformation

you

needto

run

a

nation

state.公元1749年德國(guó)阿亨瓦爾創(chuàng)造了德文詞匯“Statistik”,即“statistics”。他定義“統(tǒng)計(jì)”為治理國(guó)家所需要的信息。德文“統(tǒng)計(jì)”詞匯的出現(xiàn)27281757

Casanova

becomes

a

trustee

of,

and

may

have

hada

hand

in

devising,

the

French

nationallottery.1757年法國(guó)卡薩諾瓦成為法國(guó)國(guó)家彩票的受托人，發(fā)明了彩票。國(guó)家彩票業(yè)出現(xiàn)291761

TheRev.

Thomas

Bayes

proves

Bayes’theorem

–

the

cornerstone

of

conditional

probabilityand

the

testing

of

beliefs

and

hypotheses.1761年英國(guó)貝葉斯證明了貝葉斯定理，奠定了條件概率的基礎(chǔ)，檢驗(yàn)信念和假設(shè)。貝葉斯定理301786

William

Playfair

introduces

graphs

and

barcharts

to

show

economic

data.1786年英國(guó)愛丁堡WilliamPlayfair首次用圖表反映經(jīng)濟(jì)數(shù)據(jù)變化。經(jīng)濟(jì)數(shù)據(jù)圖表1789

Gilbert

White

and

other

clergymen-naturalistskeep

records

of

temperatures,

dates

of

firstsnowdrops

and

cuckoos,

etc;

the

data

is

later

usefulfor

study

of

climate

change.1789年英國(guó)吉爾伯特.懷特和其他牧師博物學(xué)家記錄溫度變化、首次降雪時(shí)間以及變化情況等。數(shù)據(jù)被用來研究氣候變化。31氣候統(tǒng)計(jì)1790

First

UScensus,

taken

by

men

on

horsebackdirected

by

Thomas

Jefferson,

counts

3.9

millionAmericans.1790年美國(guó)在第三任總統(tǒng)托馬斯.杰斐遜總統(tǒng)指導(dǎo)下進(jìn)行了首次人口普查，結(jié)果為390萬人口。美國(guó)首次人口普查321791

First

useof

the

word

“statistics”

in

English,by

Sir

John

Sinclair

in

his

Statistical

Account

ofScotland.1791年英國(guó)約翰.辛克萊在他“蘇格蘭統(tǒng)計(jì)賬戶”中首次使用英文詞“統(tǒng)計(jì)”Statistics.33英文“統(tǒng)計(jì)”詞匯的出現(xiàn)1805

Adrien-Marie

Legendre

introduces

the

methodof

least

squares

for

fitting

a

curve

to

a

given

set

ofobservations.1805年法國(guó)數(shù)學(xué)家勒讓德首次使用最小二乘法利用數(shù)據(jù)去擬合曲線。最小二乘法341808

Gauss,

with

contributions

from

Laplace,derivesthe

normal

distribution–

the

bell-shapedcurve

fundamentaltothe

study

of

variation

anderror.1808年德國(guó)高斯和拉普拉斯一起得到正態(tài)分布，即鐘形曲線，奠定了誤差研究的基礎(chǔ)。正態(tài)分布351833

The

British

Association

for

the

Advancementof

Science

sets

upa

statistics

section.

ThomasMalthus,

who

analysed

population

growth,

andCharles

Babbageare

members.

It

later

becomes

theRoyal

Statistical

Society.1833年，英國(guó)高等科學(xué)協(xié)會(huì)建立了統(tǒng)計(jì)分會(huì)，分析人口增長(zhǎng)的托馬斯.馬爾薩斯和查爾斯．巴貝奇都是會(huì)員，這個(gè)分會(huì)后來成為英國(guó)皇家統(tǒng)計(jì)學(xué)會(huì)。英國(guó)統(tǒng)計(jì)學(xué)會(huì)361835

Belgian

Adolphe

Quetelet’s

Treatise

on

Man

introduces

social

science

statistics

and

theconcept

of

the

“average

man”

–

his

height,

bodymass

index,

and

earnings.1835年比利時(shí)阿道夫·凱特勒在《論人及其才能的發(fā)展》中將統(tǒng)計(jì)方法用于社會(huì)科學(xué)，

并提出“平均人”的概念，討論人的身高、體重和收入等。應(yīng)用于社會(huì)科學(xué)371839:

The

American

Statistical

Association

is

formed.Alexander

Graham

Bell,

Andrew

Carnegie

andPresident

Martin

Van

Buren

will

become

members.1839年美國(guó)統(tǒng)計(jì)學(xué)會(huì)成立。亞歷山大·格雷厄姆·貝爾、德魯·卡內(nèi)基和美國(guó)總統(tǒng)馬丁?范布倫都是會(huì)員。美國(guó)統(tǒng)計(jì)學(xué)會(huì)成立381840

WilliamFarr

sets

up

the

official

system

forrecording

causes

of

death

in

England

and

Wales.This

allows

epidemics

to

be

tracked

and

diseasescompared

–the

start

of

medical

statistics.1840年，英國(guó)威廉.法爾建立了英格蘭和威爾士死亡原因的官方數(shù)據(jù)系統(tǒng)，可以追蹤研究流行病，并對(duì)疾病進(jìn)行比較研究，開創(chuàng)了醫(yī)療衛(wèi)生統(tǒng)計(jì)。醫(yī)療衛(wèi)生統(tǒng)計(jì)391849

Charles

Babbage

designs

his

“differenceengine”,

embodying

the

ideas

of

data

handlingand

the

modern

computer.

AdaLovelace,

LordByron’s

niece,

writes

the

world’s

first

computerprogram

for

it.1849年英國(guó)查爾斯．巴貝奇設(shè)計(jì)了他的“差分機(jī)”，體現(xiàn)了掌握數(shù)據(jù)和現(xiàn)代計(jì)算機(jī)的基本思想。愛達(dá)·勒芙蕾絲，拜倫勛爵的侄女，為它寫了世界上最早的計(jì)算機(jī)程序。計(jì)算機(jī)程序雛形40411854

John

Snow’s

“cholera

map”pins

down

thesource

of

an

outbreak

as

a

water

pump

in

BroadStreet,

London,

beginning

the

modern

study

ofepidemics.1854年英國(guó)約翰.斯諾利用“霍亂地圖”確認(rèn)倫敦百老匯大街的供水系統(tǒng)是疾病爆發(fā)的源頭，也是現(xiàn)代流行病學(xué)研究的源頭。開創(chuàng)流行病學(xué)研究，421859

Florence

Nightingale

uses

statistics

ofCrimean

War

casualties

to

influence

public

opinion

and

theWar

Office.

She

shows

casualties

month

by

month

on

a

circular

chart

she

devises,

the

“Nightingale

rose”,forerunner

of

the

pie

chart.

She

is

the

first

woman

member

ofthe

Royal

Statistical

Society

and

the

first

overseasmember

othe

American

Statistical

Association.1859年，南丁格爾使用克里米亞戰(zhàn)爭(zhēng)傷亡統(tǒng)計(jì)數(shù)據(jù)影響公眾意見和英國(guó)戰(zhàn)爭(zhēng)決策機(jī)構(gòu)。她將戰(zhàn)爭(zhēng)期間逐月傷亡數(shù)據(jù)用她設(shè)計(jì)的圓形圖表示出來，即“南丁格爾玫瑰”是最早的餅圖。她是英國(guó)皇家統(tǒng)計(jì)學(xué)會(huì)第一位女會(huì)員，也是美國(guó)統(tǒng)計(jì)學(xué)會(huì)第一位

海外會(huì)員。餅圖的使用431868

Minard’s

graphic

diagramof

Napoleon’sMarchon

Moscow

shows

on

one

diagram

the

distancecovered,

the

numberof

men

still

alive

ateachkilometre

of

the

march,

and

the

temperatures

theyencountered

on

the

way.1868年英國(guó)米納爾繪制了拿破侖東征莫斯科戰(zhàn)爭(zhēng)圖。圖

中繪出征程中重大戰(zhàn)役以及軍隊(duì)減員數(shù)據(jù)，從莫斯科撤退過程氣溫的變化，簡(jiǎn)潔地描描述述了了一一場(chǎng)場(chǎng)戰(zhàn)戰(zhàn)爭(zhēng)爭(zhēng)。。44統(tǒng)計(jì)圖表的妙用1877

Francis

Galton,

Darwin’s

cousin,describes

regression

to

the

mean.In

1888

he

introduces

the

concept

ofcorrelation.

At

a

“Guess

the

weightofan

Ox”

contest

in

Devon

he

describesthe“Wisdomof

Crowds”

–

that

the

average

of

manyuninformed

guesses1877年英國(guó)弗朗西斯·高爾頓，達(dá)爾文的表弟，首次描述了向

平均數(shù)的回歸。1888年他首次

使用了相關(guān)的概念。在德文郡一次“猜猜公牛的體重”的

競(jìng)賽中，他描述許多無知猜測(cè)

的平均數(shù)體現(xiàn)了“群眾的智慧”。

45回歸與相關(guān)1886

Philanthropist

Charles

Booth

begins

his

surveyof

theLondon

poor,

to

produce

his

“poverty

map

ofLondon”.

Areas

were

coloured

black,

for

thepoorest,

throughto

yellow

for

the

upper-middleclass

and

wealthy.1886年英國(guó)慈善家查爾斯·布斯開始在倫敦進(jìn)行貧困調(diào)查并繪制了“貧困地圖”。地圖上用黑色表示最貧窮地區(qū)，逐漸過渡到黃色代表中上層和富裕地區(qū)。貧困地圖461894

Karl

Pearson

introduces

the

term

“standarddeviation”.

If

errors

are

normally

distributed,

68%

osamples

will

lie

within

one

standard

deviation

of

themean.

Later

he

develops

chi-squared

tests

forwhether

two

variables

are

independent

of

each

other.1894年英國(guó)卡爾.皮爾遜首次使用了“標(biāo)準(zhǔn)差”的術(shù)語(yǔ)。如果誤差是正態(tài)分布的，68%的樣本會(huì)落在均值附近正負(fù)一個(gè)標(biāo)準(zhǔn)差之內(nèi)。不久，他又提出卡方檢驗(yàn)，用來檢驗(yàn)兩個(gè)統(tǒng)計(jì)量是否獨(dú)立。標(biāo)準(zhǔn)差47481898

Von

Bortkiewicz’s

data

on

deaths

of

soldiersin

the

Prussian

army

from

horse

kicks

shows

thatapparently

rare

events

follow

a

predictable

pattern,the

Poisson

distribution1898年，德國(guó)范.鮑特凱維茲發(fā)現(xiàn)普魯士士兵被戰(zhàn)馬踢死的數(shù)據(jù)是明顯的稀有事件，可以用泊松分布進(jìn)行預(yù)測(cè)。泊松分布三、現(xiàn)代發(fā)展（20世紀(jì)初至今）49501900

Louis

Bachelier

shows

that

fluctuations

instock

market

prices

behave

inthe

same

way

as

therandom

Brownian

motion

of

molecules

–

the

startof

financial

mathematics.1900年，法國(guó)數(shù)學(xué)家巴施里葉（Louis

Bachelier）首先發(fā)現(xiàn)股票市場(chǎng)價(jià)格的波動(dòng)與分子隨機(jī)布朗運(yùn)動(dòng)一致，開創(chuàng)了金融數(shù)學(xué)。金融數(shù)學(xué)1908

William

Sealy

Gossett,

chief

brewer

forGuinness

in

Dublin,

describes

the

t-test.

It

uses

asmall

numberof

samplesto

ensure

that

every

brewtastes

equally

good.1908年英國(guó)都柏吉尼斯啤酒廠的首席釀酒師威廉.戈塞特提出了t檢驗(yàn)。使用小樣本確保每一桶啤酒都是一樣

的高質(zhì)量。小樣本t統(tǒng)計(jì)量511911

Herman

Hollerith,

inventorof

punch-carddevices

usedto

analyse

data

in

US

censuses,merges

his

company

toform

what

will

becomeIBM,

pioneers

of

machinesto

handle

business

dataandof

early

computers.1911年在美國(guó)人口普查中使用打孔機(jī)分析數(shù)據(jù)的赫爾曼.霍爾瑞斯將收購(gòu)的公司與自己公司合并形成IBM，是使用機(jī)器處理數(shù)據(jù)和早期計(jì)算機(jī)的先驅(qū)。計(jì)算機(jī)處理數(shù)據(jù)先驅(qū)52。531916

During

the

First

World

Warcar

designerFrederick

Lanchester

develops

statistical

laws

topredict

the

outcomes

of

aerial

battles:

if

you

doubletheir

size

land

armies

are

only

twice

as

strong,

butair

forces

are

four

times

as

powerful.1916年第一次世界大戰(zhàn)期間，英國(guó)汽車的設(shè)計(jì)者蘭徹斯特用統(tǒng)計(jì)法則預(yù)測(cè)空戰(zhàn)結(jié)果如果陸軍軍隊(duì)數(shù)量是對(duì)手兩倍的話，其空軍戰(zhàn)斗力將是對(duì)手的四倍。第一次世界大戰(zhàn)中統(tǒng)計(jì)的應(yīng)用541924

Walter

Shewhartinvents

the

control

chart

toaid

industrial

production

and

management1924年美國(guó)貝爾實(shí)驗(yàn)室的沃爾特.休哈特發(fā)明了控制圖，極大地提高了工業(yè)生產(chǎn)和管理水平。質(zhì)量控制圖551935

George

Zipf

finds

that

many

phenomena

–

riverlengths,

city

populations

–

obey

a

power

law

so

thatthe

largest

is

twice

the

size

of

the

second

largest,three

times

the

size

of

the

third,

and

so

on.1935年美國(guó)語(yǔ)言學(xué)家約翰.齊普夫發(fā)現(xiàn)許多現(xiàn)象，如河流長(zhǎng)度、城市人口數(shù)、英文詞匯使用頻率等都遵從一條定律即出現(xiàn)最多的是出現(xiàn)第二多的兩倍，是出現(xiàn)第三多的三倍，等等，被稱為齊普夫定律，也就是我們常說的“二八原則”。齊普夫定律：“二八”原則561935

R.

A.

Fisher

revolutionises

modern

statistics.His

Design

of

Experiments

gives

waysof

decidingwhich

results

of

scientific

experiments

aresignificant

and

whichare

not.1935年英國(guó)費(fèi)雪對(duì)現(xiàn)代統(tǒng)計(jì)學(xué)作出了歷史性的貢獻(xiàn)。他

的試驗(yàn)設(shè)計(jì)方法能夠確定哪些科學(xué)試驗(yàn)結(jié)果是顯著的，哪些不是?，F(xiàn)代試驗(yàn)設(shè)計(jì)571937

Jerzy

Neyman

introduces

confidence

intervalsin

statistical

testing.

His

workleads

to

modernscientific

sampling.1937年內(nèi)曼在統(tǒng)計(jì)檢驗(yàn)中給出了置信區(qū)間，他的成果開創(chuàng)了現(xiàn)代科學(xué)抽樣理論。置信區(qū)間581940-45

Alan

Turing

at

Bletchley

Park

cracks

theGerman

wartime

Enigma

code,using

advancedBayesian

statistics

and

Colossus,thefirstprogrammable

electronic

computer.1940-45年英國(guó)數(shù)學(xué)家阿蘭.圖靈在布萊切利園破解德軍戰(zhàn)爭(zhēng)中的密碼，他使用高等貝葉斯統(tǒng)計(jì)，并研制了巨人計(jì)算機(jī)，第一臺(tái)編程的電子計(jì)算機(jī)。首臺(tái)編程計(jì)算機(jī)591944

The

German

tank

problem:

the

Allies

desperately

needto

know

how

many

Panther

tanks

they

will

face

in

France

onD-Day.

Statistical

analysis

of

the

serial

numbers

ongearboxes

from

captured

tanks

indicates

how

many

of

eachare

being

produced.

Statisticians

predict

270

a

month;reports

from

intelligence

sources

predict

many

fewer.

Thetotal

turned

out

to

be

276.

Statistics

had

outperformed

spies1944年德軍坦克問題：盟軍急切地想知道在法國(guó)將面對(duì)

多少豹式坦克。破解了被繳獲坦克變速箱上的生產(chǎn)序列號(hào)利用統(tǒng)計(jì)方法預(yù)測(cè)德軍每月增加270輛，諜報(bào)人員預(yù)測(cè)比270少得多，最后實(shí)際數(shù)字是

276輛，統(tǒng)計(jì)預(yù)測(cè)勝過諜報(bào)工作。第二次世界大戰(zhàn)中統(tǒng)計(jì)的應(yīng)用1946

Cox’s

theoremderives

the

axioms

ofprobability

from

simple

logical

assumptions.1946年美國(guó)考克斯利用簡(jiǎn)單邏輯假設(shè)推出了概率論公理。概率論公理60611948

Claude

Shannon

introduces

informationtheoryand

the“bit”–fundamental

tothe

digital

age.1948年美國(guó)科學(xué)家克勞德.香農(nóng)提出了信息論和“比特”的概念，開創(chuàng)了數(shù)字時(shí)代。信息論與比特1948-53

The

Kinsey

Report

gathers

objective

data

onhumansexual

behaviour.

A

large-scale

survey

of5000

men

and,

later,

5000

women,

it

causes

outrage.1948-53年美國(guó)金 報(bào)告收集了人類性行為的客觀數(shù)據(jù)，先是對(duì)5000男性進(jìn)行了調(diào)查，接著又對(duì)5000女性進(jìn)行調(diào)查，報(bào)告公開后引起社會(huì)巨大憤怒和反響。性學(xué)研究62631950

Richard

Doll

and

Bradford

Hill

establish

the

linkbetween

cigarette

smoking

and

lung

cancer.

Despitefierce

opposition

the

result

is

conclusively

proved,to

hugepublic

health

benefit.1950年英國(guó)理查德.多爾和布萊德?！は栄芯苛顺闊熀头伟┑年P(guān)系。他們頂住反對(duì)意見的壓力，最終給出了研究的結(jié)果，對(duì)公眾健康有益。抽煙與肺癌研究1950s

GenichiTaguchi’s

statistical

methods

toimprove

the

quality

ofautomobile

and

electronicscomponents

revolutionise

Japanese

industry,

whichfar

overtakes

western

European

rivals.20世紀(jì)50年代日本田口玄一利用統(tǒng)計(jì)方法改善汽車和電

子產(chǎn)品給日本工業(yè)界帶來革命，使得日本遠(yuǎn)遠(yuǎn)超過歐美競(jìng)爭(zhēng)者的質(zhì)量。田口的試驗(yàn)設(shè)計(jì)64651958

The

Kaplan–Meier

estimator

gives

doctors

asimple

statistical

wayof

judging

which

treatmentswork

best.

It

has

saved

millions

oflives.1958年美國(guó)Kaplan–Meier生存分析估計(jì)方法使得醫(yī)生可以用簡(jiǎn)單的統(tǒng)計(jì)方法判斷治療方案的有效性，解救了數(shù)百萬生命。生存分析1972

David

Cox’s

proportional

hazard

model

andthe

concept

of

partial

likelihood.1972年英國(guó)大衛(wèi).考克斯使用比例風(fēng)險(xiǎn)模型和偏似然函數(shù)概念。比例風(fēng)險(xiǎn)模型與偏似然函數(shù)661977

John

Tukey

introduces

the

box-plot

orbox-and-whisker

diagram,

which

shows

thequartiles,medians

and

spread

of

data

in

a

singleimage.1977年美國(guó)約翰.圖基介紹了箱線圖和莖葉圖，利用數(shù)據(jù)的四分位數(shù)、中位數(shù)和散布等簡(jiǎn)單直觀表示數(shù)據(jù)特征。探索性數(shù)據(jù)分析671979

Bradley

Efron

introduces

bootstrapping,

asimplewayto

estimate

the

distribution

of

almostany

sample

of

data.1979年美國(guó)斯坦福大學(xué)布拉德利.艾佛容提出了自助法，一種可以估計(jì)任意樣本數(shù)據(jù)抽樣分布的簡(jiǎn)單方法。自助法681982

EdwardTufte

self-publishes

The

VisualDisplay

of

Quantitative

Information,

setting

newstandards

for

graphic

visualisation

of

data.1982年美國(guó)耶魯大學(xué)愛德華·塔夫特自己印制了“數(shù)量信息的可視化”，建立了數(shù)據(jù)圖形可視化新標(biāo)準(zhǔn)。圖形可視化691988

Margaret

Thatcher

becomes

the

first

worldleader

to

call

for

action

onclimate

change.1988年英國(guó)首相瑪格麗特·撒切爾號(hào)召采取行動(dòng)面對(duì)氣候變化，成為世界上首位領(lǐng)導(dǎo)人。應(yīng)對(duì)氣候變化701993

The

statistical

programming

language

“R”

isreleased,

nowa

standard

statistical

tool.1993年統(tǒng)計(jì)“R”語(yǔ)言發(fā)布使用，現(xiàn)在成為一種標(biāo)準(zhǔn)的統(tǒng)計(jì)工具。“R”語(yǔ)言是20世紀(jì)80年代從美國(guó)“S”語(yǔ)言基礎(chǔ)上發(fā)展起來的?！癛”語(yǔ)言711997

The

term“Big

Data”first

appears

in

print.1997年“大數(shù)據(jù)”一詞首次見諸報(bào)端?！按髷?shù)據(jù)”首次出現(xiàn)722002

The

amountof

informationstoreddigitallysurpasses

non-digital.2002年以數(shù)字形式存儲(chǔ)的信息首次超過非數(shù)字信息?！皵?shù)字”數(shù)據(jù)首次為主73742002

Paul

DePodesta

uses

statistics

–“sabermetrics”–

to

transform

the

fortunes

of

the

Oakland

Athleticsbaseball

team;thefilm

Moneyball

tellsthestory.2002年美國(guó)保爾.波戴斯塔應(yīng)用統(tǒng)計(jì)方法“棒球數(shù)據(jù)分析”改變了奧克蘭運(yùn)動(dòng)家隊(duì)的命運(yùn)，電影“點(diǎn)球成金”講述了這一故事。體育中應(yīng)用752004

Launchof

Significance

magazine.2004年“Significance”雜志創(chuàng)刊?！癝ignificance”雜志創(chuàng)刊2008

Hal

Varian,

chief

economist

at

Google,

saysthat

statistics

will

be

“the

sexy

professionof

thenext

ten

years”.2008年美國(guó)谷歌首席經(jīng)濟(jì)學(xué)家預(yù)測(cè)統(tǒng)計(jì)學(xué)將成為未來十年最性感的職業(yè)。對(duì)統(tǒng)計(jì)學(xué)的預(yù)測(cè)762012

The

Large

Hadron

Collider

confirmsexistenceof

a

Higgs

boson-like

particle

with

probability

of

fivstandarddeviations–

around

one

chance

in

3.5million

that

all

they

are

seeing

is

coincidence.2012年歐洲大型電子對(duì)撞機(jī)確認(rèn)希格斯玻色子粒子的存在，發(fā)生的概率是5個(gè)標(biāo)準(zhǔn)差外的小概率，即350萬分之一的概率。希格斯玻色子粒子772012

Nate

Silver,

statistician,

successfully

predictsthe

result

in

all

50

states

in

the

US

Presidentialelection.

He

becomesa

media

star.2012年美國(guó)統(tǒng)計(jì)學(xué)家內(nèi)特.席爾瓦成功地預(yù)測(cè)了美國(guó)所有50個(gè)州總統(tǒng)大選的結(jié)果，成為媒體明星。美國(guó)大選成功預(yù)測(cè)7879早期源頭：公元前5世紀(jì)--