David A. Freedman (statistician)

from Wikipedia, the free encyclopedia
David A. Freedman

David Amiel Freedman (born March 5, 1938 in Montreal , Canada , † October 17, 2008 in Berkeley , California ) was a Canadian-American applied and mathematical statistician.

Life

Freedman studied at McGill University with a bachelor's degree in 1958 and at Princeton University with a master's degree in 1959 and a doctorate with William Feller in 1960 (Mixtures of stochastic processes). As a post-doctoral student , he was a Canada Council Fellow at Imperial College London in 1960/61 . In 1961 he became a lecturer and in 1962 a professor at the University of California, Berkeley . In 1990 he was a Miller Fellow at Berkeley.

He was visiting professor in Athens, Caracas (1970/71), Jerusalem (1968/69), Kuwait (1981), London and Mexico City (1973).

Freedman was married twice and had two children.

plant

Freedman first studied mathematical statistics ( Bayesian learning ) and then became a leading international scientist in applied statistics. He advised banks, commercial enterprises, state and other organizations on the application of statistics, for example in the right to vote, on the question of discrimination in the workplace, fair wages, the trajectory of golf balls, mad cow disease, ecological questions, railway taxation, errors in price scanners, Multiple signatures for petitions, breast cancer, earthquake probability. With his colleague Kenneth Wachter, he testified before the US Congress regarding statistical corrections in the 1980 and 1990 censuses. The US Supreme Court followed their recommendations in the spirit of the US Department of Commerce. His introduction to statistics for lawyers found widespread use in legal circles in the United States. It is published at the Federal Judicial Center of the US Justice.

He investigated how statistical methods behave when the underlying assumptions are wrong and characterized cases in which the methods still worked well and cases in which this was not the case regardless of the quality of the data. For example, he demonstrated in 1965 ( On the asymptotic behavior of Bayes estimates in the discrete case ) that the method of Bayesian learning ( Bernstein-von-Mises' theorem ) breaks down almost everywhere in the case of countably infinitely large sampling populations. In contrast, the method works for a finite sampling population ( Joseph L. Doob 1948). With Persi Diaconis in 1986 he showed that this result also has an effect on high-dimensional problems.

With Diaconis he gave an estimate of the class width in histograms in 1981 (rule of Freedman and Diaconis, see histogram # construction of a histogram ).

His Introduction to Statistics first appeared in 1978 and had four editions. It has been translated into Chinese, Hungarian and Spanish.

In mathematical statistics, he also dealt with Markov chains , Brownian motion and diffusion, ergodic theory , martingales .

Memberships and honors

Freedman was a fellow of the Institute of Mathematical Statistics , the American Statistical Association and, since 1991, the American Academy of Arts and Sciences . In 2003 he received the John J. Carty Award . From 1964 to 1966 he was a Sloan Research Fellow .

Fonts

  • On the asymptotic behavior of Bayes estimates in the discrete case, 2 parts, Annals of Mathematical Statistics, Volume 34, 1963, pp. 1386-1403, Volume 36, 1965, pp. 454-456.
  • Markov Chains, Holden-Day 1971, Springer 1983
  • Brownian Motion and Diffusion, Holden-Day 1971, Springer 1983
  • Statistical models: theory and practice, Cambridge UP 2005
  • with Robert Pisani, Roger Purves: Statistics, New York: WW Norton, 1978, 4th edition 2007
  • Statistical Models and Causal Inference: A Dialogue with the Social Sciences, Cambridge UP 2009 (Editors David Collier, Jasjeet Sekhon, Philip B Stark)
  • with David Kaye: Reference Guide on Statistics, in: Reference Manual on Scientific Evidence, Online, National Academy , Federal Judicial Center
  • with Diaconis: On the consistency of Bayes estimates , Annals of Statistics, Volume 14, 1986, pp. 1-26, Project Euclid
  • On the Bernstein von Mises theorem with infinite dimensional parameters , Annals of Statistics, Volume 27, 1999, pp. 1119-1140

Web links

Individual evidence

  1. ↑ David A. Freedman in the Mathematics Genealogy Project (English)Template: MathGenealogyProject / Maintenance / id used
  2. ↑ Career data Pamela Kalte u. a., American Men and Women of Science, Thomson Gale, 2004
  3. ^ Diaconis, Freedman: On the histogram as a density estimator: L2 theory, Probability Theory and Related Fields, Volume 57, 1981, pp. 453-476.
  4. ↑ Book of Members 1780 – present, Chapter F. (PDF; 815 kB) In: American Academy of Arts and Sciences (amacad.org). Accessed October 30, 2018 .
personal data
SURNAME Freedman, David A.
ALTERNATIVE NAMES Freedman, David Amiel (full name)
BRIEF DESCRIPTION Canadian-American applied and mathematical statistician
DATE OF BIRTH March 5, 1938
PLACE OF BIRTH Montreal , Canada
DATE OF DEATH October 17, 2008
Place of death Berkeley , California