60

What are the most important statisticians, and what is it that made them famous?
(Reply just one scientist per answer please.)

Nick Stauner
  • 11,558
  • 5
  • 47
  • 105
mariana soffer
  • 1,091
  • 2
  • 15
  • 18

46 Answers46

91

Ronald Fisher for his fundamental contributions to the way we analyze data, whether it be the analysis of variance framework, maximum likelihood, permutation tests, or any number of other ground-breaking discoveries.

Christopher Aden
  • 1,775
  • 4
  • 24
  • 43
  • 4
    It's worth noting that Fisher is equally famous for his work as a biologist (evolutionary biology and agricultural science) as he is for his statistical work. – Michael Lew Dec 04 '10 at 06:49
  • 14
    *Even today, I occasionally meet geneticists who ask me whether it is true that the great geneticist R. A. Fisher was also an important statistician* -- Leonard Savage (Annals of Statistics, 1976 http://www.jstor.org/stable/2958221). – onestop Dec 04 '10 at 08:06
  • 2
    @Michael Lew: He certainly is. He managed to get over the supposed disconnection between Mendelian genetics and Darwinian evolution, among other accomplishments. http://digital.library.adelaide.edu.au/coll/special//fisher/9.pdf – Christopher Aden Dec 05 '10 at 00:18
  • 6
    If you want to ask a tough question, ask who the second most famous statistician is. There's no doubt that Fisher is # 1. You may not like the man or some of his ideas, but he is undoubtedly the creator of Statistics as we know it today. – Carlos Accioly Dec 06 '10 at 22:41
  • 3
    @Carlos "undoubtedly the creator" are you a member of FA (Fisher's Adorators) ? Anyway Gauss introduced least square almost a hundred year before Fisher was born, and Fisher as well as Gauss were inspired by a lot of other very inspired people... it is a long and laborious story and unfortunatly for those who like THX surround movies I don't really think it has its Guru. – robin girard Dec 15 '10 at 07:30
54

John Tukey for Fast Fourier Transforms, exploratory data analysis (EDA), box plots, projection pursuit, jackknife (along with Quenouille). Coined the words "software" and "bit".

Neil McGuigan
  • 9,292
  • 13
  • 54
  • 62
53

Reverend Thomas Bayes for discovering Bayes' theorem

Neil McGuigan
  • 9,292
  • 13
  • 54
  • 62
  • But was he _really_ a statistician? – vqv Dec 19 '10 at 22:27
  • 2
    That's a tough one...when did statistics become a real field? Many of the fathers of stats were not statisticians. – Neil McGuigan Dec 20 '10 at 18:23
  • 1
    I know my choice is kind of arbitrary, because many are important but this is my favourite one, and his method allowed me to do lots of things. – mariana soffer Dec 24 '10 at 18:35
  • 21
    That this was selected indicates a biased prior. – Iterator Aug 06 '11 at 00:20
  • 1
    Does one *discover* a theorem? Shouldn't it be *postulating* or *theorizing*? – nico Dec 04 '11 at 13:21
  • 7
    Sigh. I would have hoped that one could be an enthusiast of the Bayesian approach without actually believing that Bayes the man was the greatest statistician of all time. Shouldn't we honor someone who contributed important work to its development, like, e.g., [Jaynes](http://en.wikipedia.org/wiki/Edwin_Thompson_Jaynes)? – gung - Reinstate Monica Jan 07 '12 at 03:26
  • 1
    As I understand it, Thomas Bayes (ca 1701 to 1761) never published the theorem that has been attributed to him. Richard Price, from Bayes writings, introduced it to the Royal Society two years after Baye's death which led to its posthumous publication in 1764. There it essentially lay dormant until the principle was rediscovered, developed and applied by Pierre Simon Laplace around 1774. But Laplace's formulation was so different that one "can discount that he had seen Baye's essay" (Stigler, The History of Statistics, Harvard University Press, 1986, 2003). – AsymLabs Jan 11 '18 at 14:41
49

Karl Pearson for his work on mathematical statistics. Pearson correlation, Chi-square test, and principal components analysis are just a few of the incredibly important ideas that stem from his works.

Dason
  • 430
  • 5
  • 9
47

Carl Gauss for least squares estimation.

Neil McGuigan
  • 9,292
  • 13
  • 54
  • 62
45

William Sealy Gosset for Student's t-distribution and the statistically-driven improvement of beer.

IanS
  • 126
  • 2
  • 4
  • 6
    Maybe not the most famous statistician, but when you put it like that, definitely the most important! ;o) – Dikran Marsupial Dec 05 '10 at 00:13
  • 6
    @ IanS - to be precise, he improved the quality of Guiness, which is a stout, not a beer (in case you hadn't noticed, I'm Irish). – richiemorrisroe Sep 23 '11 at 10:34
  • 1
    A stout is a beer, by every reasonable definition of beer. Roasting barley does not transform the beverage into an entirely different type of beverage. – Behacad Oct 18 '13 at 14:31
41

Bradley Efron for the Bootstrap - one of the most useful techniques in computational statistics.

Dikran Marsupial
  • 46,962
  • 5
  • 121
  • 178
39

Andrey Nikolayevich Kolmogorov, for putting probability theory on a rigorous mathematical footing. While he was a mathematician, not a statistician, undoubtedly his work is important in many branches of statistics.

LeelaSella
  • 1,770
  • 3
  • 24
  • 42
  • 5
    +1 for Kolmogorov. Without him, a rigorous treatment of most fundamental statistical concepts would not be possible and therefore statistics not as reliable as they are now. – Thilo Dec 04 '11 at 13:42
  • 2
    Amen. His Probability Axioms combined with Measure Theory (Émile Borel, Henri Lebesgue, Johann Radon and Maurice Fréchet) could be considered the most powerful analytical technique in applied mathematics, in my opinion. – AsymLabs Oct 19 '13 at 05:46
36

Pierre-Simon Laplace for work on fundamentals of (Bayesian) probability.

Dikran Marsupial
  • 46,962
  • 5
  • 121
  • 178
30

George Box for his work on time series, designed experiments and elucidating the iterative nature of scientific discovery (proposing and testing models).

Jeromy Anglim
  • 42,044
  • 23
  • 146
  • 250
29

Francis Galton for discovering statistical correlation and promoting regression.

Neil McGuigan
  • 9,292
  • 13
  • 54
  • 62
  • 3
    +1 for Francis Galton played a very important role in giving importance to the concept of correlation. However, I found a bit strong the formulation "creating correlation". I would quote Galton itself: << "Co-relation or correlation of structure" is a phrase much used in biology, and not least in that branch of it which refers to heridity, and the idea is even more frequently present than the phrase; but I am not aware of any attempt to define it clearly >> In : "Co-relation and their Measurment" (see here http://galton.org/galton/index.html) – robin girard Dec 15 '10 at 09:19
  • @robin: You could say that Galton did not create the idea of correlation, but did create the first statistical representation of it. Google "francis galton's account of the invention of correlation" – Neil McGuigan Dec 15 '10 at 19:03
  • 3
    -1 In1884,Galton wrote:"Jews are specialized for a parasitical existence upon other nations,and that there is need of evidence that theyare capable of fulfilling thevaried duties ofa civilized nation by themselves."Karl Pearson,Galton's disciple and biographer,echoed this opinion 40years later during his attempt to prove the undesirability of Jewish immigration into Britain:"for such men as religion,social habits,or language keep asa caste apart,there should be no place.they will not be absorbed by,and at the same time strengthenthe existing population;they will develop into a parasitic race" – IrishStat May 22 '11 at 16:21
  • 5
    @IrishStat. Yup, a lot of the early statisticians were eugenists, including Pearson and Fisher. But, they were still great statisticians. – Neil McGuigan May 22 '11 at 17:26
  • Can you elaborate on Fisher in this regard ? – IrishStat May 22 '11 at 21:27
  • 5
    @IrishStat: Fisher was the first president of the Cambridge University Eugenics Society, wrote for *Eugenics Review* and *Annals of Eugenics* from 1914 to 1947, and was appointed professor of eugenics at University College London in 1933. – Henry Sep 23 '11 at 00:27
28

Andrey Markov for stochastic processes and markov chains.

Neil McGuigan
  • 9,292
  • 13
  • 54
  • 62
  • was he a statistician? I mean, did he wrote things about estimation? – robin girard Dec 13 '10 at 18:26
  • 2
    he was a mathematician that was very important to statistics. Half of the men on here lived before statistics was an "official" field of study . – Neil McGuigan Dec 15 '10 at 08:48
  • good point ! I guess I meant that even before it was official, some mathematicians (like Gauss) concidered statistical problems while I don't really know if Markov did concider statistical problem himself. – robin girard Dec 15 '10 at 18:01
26

Jerzy Neyman and Egon Pearson for work on experimental design, hypothesis testing, confidence intervals, and the Neyman-Pearson lemma.

IanS
  • 126
  • 2
  • 4
25

Leo Breiman for CART, bagging, and random forests.

Neil McGuigan
  • 9,292
  • 13
  • 54
  • 62
24

How has Sir David Roxbee Cox not been mentioned yet?

Some feats: Cox proportional hazards models, experimental design, he did a lot of work on stochastic processes and binary data. He also advised many students who went on to do great work (Hinkley, McCullagh, Little, Atkinson, etc.)

And the man was knighted!

  • +1 for Sir D.R. Cox, a great man indeed. He and co-autors wrote a number of excellent books which I believe inspired many researchers and applied statisticians around the world, and are major contributions to modern statistics. – Yves Dec 05 '14 at 17:23
23

Harold Jeffreys for revival of Bayesian interpretation of probability.

Dikran Marsupial
  • 46,962
  • 5
  • 121
  • 178
21

Edwin Thompson Jaynes for work on objective Bayesian methods, particularly MaxEnt and transformation groups.

Dikran Marsupial
  • 46,962
  • 5
  • 121
  • 178
  • 4
    For me, he is the best statistical thinker ever. And also one of the most interesting and distinctive writing styles I've seen. Never seen anyone use the word "common sense" like he does! – probabilityislogic Mar 12 '11 at 14:29
  • 1
    Jaynes' book "Probability Theory" has no equal, in my opinion, for one who wishes a deep understanding the subject in practice. – AsymLabs Jan 09 '18 at 20:07
21

Blaise Pascal and Pierre de Fermat for creating the theory of probability and inventing the idea of expected value (1654) in order to solve a problem grounded in statistical observations (from gambling).

whuber
  • 281,159
  • 54
  • 637
  • 1,101
20

C.R. Rao for the Rao–Blackwell theorem and the Cramer-Rao bound.

Innuo
  • 1,418
  • 10
  • 11
20

Florence Nightingale for being "a true pioneer in the graphical representation of statistics" and developing the polar area diagram. Yes, that Florence Nightingale!

Jeromy Anglim
  • 42,044
  • 23
  • 146
  • 250
  • 4
    Nightingale did excellent work, but her originality is often exaggerated. Polar diagrams were used by several people before her, earlier in the 19th century. – Nick Cox May 27 '13 at 10:30
  • 3
    This is what I explain to nursing students, to demonstrate to them that they should be interested in statistics. Doesn't work. – Jeremy Miles Oct 18 '13 at 20:54
13

Roderick Little and Donald Rubin for the contributions in Missing Data Analysis.

Biostat
  • 1,791
  • 2
  • 19
  • 21
  • 4
    Don Rubin also had a hand in the EM algorithm and the development of causal inference. – guy Aug 02 '12 at 15:11
11

W. Edwards Deming for promoting statistical process control

Neil McGuigan
  • 9,292
  • 13
  • 54
  • 62
10

George Dantzig for the Simplex Method, and for being the student who mistook two open statistics problems that Neyman had written on the board for homework problems, and in his "ignorance" solving them. I'd vote for him just for the story.

Wayne
  • 19,981
  • 4
  • 50
  • 99
  • 2
    This story is identical to one told about John Milnor. In Milnor's case there's at least one paper (co-authored with his professor when Milnor was still an undergraduate) to give the story credence. Have you ever found a reference to the paper(s) Dantzig wrote giving his solutions? – whuber Dec 13 '10 at 23:17
  • 1
    According to Wikipedia, the first paper was with/by Neyman and the second result was enough to get Dantzig a co-authorship with Abraham Wald. No actual references given, though. – Wayne Dec 14 '10 at 04:09
  • 2
    Thanks. It is interesting how such a story can recur and how it might actually be true in both cases! – whuber Dec 15 '10 at 17:39
  • 2
    @whuber Yes, it is! My skept-o-meter says that the details of multiple stories may not be quite right, though. Perhaps both of them solved open problems, but maybe one of them knew the problem was open and the other mistook it for a homework problem. The stories would match in terms of "naive young student solves open problem, not having realized how hard it really was", but when the story became legend and someone told it as "some student ...", someone else chimed in, "oh, that was Dantzig", or "oh, that sounds like Milnor" and the stories converged. – Wayne Dec 16 '10 at 15:44
8

Abraham Wald (1902-1950) for introducing the concept of Wald-tests and for his fundamental work on statistical decision theory.

Michael M
  • 10,553
  • 5
  • 27
  • 43
8

Samuel S. Wilks was a leader in the development of mathematical statistics. He developed the theorem on the distribution of the likelihood ratio, a fundamental result that is used in a wide variety of situations.

He also helped found the Princeton statistics department, where he was Fred Mosteller's advisor, among others, and has a prestigious ASA award named after him.

Aaron left Stack Overflow
  • 6,808
  • 29
  • 51
6

David Donoho development of multiscale ideas in statistics, and a lot of theoretically justified while practically very efficient ideas in very high dimensional statistics, CHA: computational harmonic analysis,...

whuber
  • 281,159
  • 54
  • 637
  • 1,101
robin girard
  • 6,335
  • 6
  • 46
  • 60
6

Emanuel Parzen for kernel density estimation and reproducing kernel Hilbert space theory for stochastic processes.

Memming
  • 1,570
  • 12
  • 23
robin girard
  • 6,335
  • 6
  • 46
  • 60
6

Lucien Le Cam for his contribution to mathematical statistics. (maybe Local asymptotic normality and contiguity made him famous)

robin girard
  • 6,335
  • 6
  • 46
  • 60
6

Leland Wilkinson for his contribution to statistical graphics.

Andrej
  • 2,131
  • 2
  • 18
  • 26
6

Adolphe Quetelet for his work on the "average man", and for pioneering the use of statistics in the social sciences. Before him, statistics were largely confined to the physical sciences (astronomy, in particular).

5

Robert Gentleman and Ross Ihaka for being the two initiators of and later main contributors of R, see https://cran.r-project.org/doc/html/interface98-paper/paper_1.html

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
5

John Nelder, for providing us the now omnipresent generalized linear model framework. By his approach of unifying various standard statistical models and its estimation method, the iteratively reweighted least squares method for ML, he gave us tools that we are using now in almost all applied and theoretical concepts that are related to exponential family models. Not to mention his contributions to optimization as the superb Nelder-Mead-Algorithm.

Michael L.
  • 1,318
  • 6
  • 16
5

It's very difficult to add to the constellation of stars that are already listed, but for interest purposes I will throw in the improbable polymath John Maynard Keynes who many would not realize published A Treatise on Probability (1921) that can be downloaded here; and whose work was quoted frequently by Harold Jeffreys (1939).

Keynes by all accounts helped to bring forward Bayesian statistics and in his treatise considered the most important principle to be the Principle of Indifference.

According to Wikipedia, The "Principle of insufficient reason" was renamed the "Principle of Indifference" by the economist John Maynard Keynes (1921), who was careful to note that it applies only when there is no knowledge indicating unequal probabilities.

AsymLabs
  • 251
  • 2
  • 8
4

William Cleveland, for either coining or popularising the term 'data science', but more importantly for contributions to data visualisation including the popular 'dot plot', and contributions to nonparametric regression such as loess smoothing.

Robert de Graaf
  • 1,359
  • 10
  • 14
3

Brian Ripley for contributions to spatial statistics, neural network from statistician point of view, and a lot of other contributions, not the least of them being a main contributor to the success of R!

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
3

Teuvo Kohonen for invention of the Self-Organizing-Map (SOM).

mlwida
  • 9,922
  • 2
  • 45
  • 74
  • 2
    and maybe also Bishop, Svensen and Williams for putting SOM on a clean probabilistic footing via the Generative Topographic Mapping. (And I didn't down vote anything either...) – conjugateprior Dec 23 '10 at 23:13
  • self-organizing precursors of Kohonen included Alan Turing (1952), Willshaw, Bunenman and Longuet-Higgens (1960s) and von der Malsburg (1973). (Didn't down vote anything either...) – David M W Powers Mar 29 '16 at 11:35
2

Vladimir Vapnik:

For his fundamental contributions to our understanding of machine learning, which allows computers to classify new data based on statistical models derived from earlier examples, and for his invention of widely used machine learning techniques.

MMut
  • 11
  • 2
2

I'd like to also add William Gemmell Cochran who is well known for establishing (or directing studies at) some of the preeminent statistics departments such as at Iowa State University, Harvard, Cambridge, that educated hundreds of future statisticians. His methods on Survey Sampling Techniques and Design and Analysis of Experiments are still widely used today.

StatsStudent
  • 10,205
  • 4
  • 37
  • 68
2

Bill James for his work in statistics that evaluate MLB player performance. His work spawned the term Sabermetics. He has created numerous statistics that can be found throughout the baseball world. His ideas stem from how to capture a player's overall impact on a game through run production (offense) and runs saved (defense). His work has led to less emphaisis on statistics that have low correlation to run production (batting average) and more on OPS (on-base + slugging). He works as an advisor to the Boston Red Sox and is credited to the World Series Championships in 2004 and 2007. His work has influenced the book and upcoming feature film Moneyball.

1

Harald Cramér was a swedish statistician, mathematician and actuary. John Kingman described him as "one of the giants of statistical theory".

Among his first works was application of probability in number theory. His 1946 book Mathematical Methods of Statistics was very influential, in showing a place for mathematical rigour in statistics. Most people will have heard about the Cramér-Rao inequality.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
1

If you judge based on the number of times their work is cited, E. L. Kaplan and Paul Meier since their 1958 paper "Nonparametric Estimation from Incomplete Observations" is widely regarded as the most cited paper in statistics.

StatsStudent
  • 10,205
  • 4
  • 37
  • 68
1

Lotfi A. Zadeh, who not only coined the terms "fuzzy set" and "fuzzy logic", but invented many of the accompanying fuzzy statistics (like the sigma count, for instance).

SQLServerSteve
  • 1,121
  • 1
  • 13
  • 34
1

Joseph Hilbe (1944-), first president of the International Astrostatistics Association and author of over 10 books on statistical modeling, including popular texts on count models, logistic regression, generalized estimating equations (GEE), generalized linear models, and statistical methodology. Hilbe is an emeritus professor at the University of Hawaii and adjunct professor of statistics at Arizona State University.

Michael Brown
  • 1
  • 1
1

John Kingman for Coalescent theory and his work on completely random measures

Christos
  • 490
  • 2
  • 13
0

Jerome Harold Friedman for his contributions on CART (jointly with Leo Breiman in 1984), gradient boosting machines and partial dependence plots. Plus being co-author of the book (ESL)...

Michael M
  • 10,553
  • 5
  • 27
  • 43
0

Irving John Good for contributions, among others, to Bayesian Statistics. He learnt probability from Turing at Bletchley Park. Some of his ideas was mentioned at What is the role of the logarithm in Shannon's entropy? which has references.

He was also a consultant for Kubrick at the famous film 2001: A Space Odyssey.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467