“We are very empirical,” said James R. Thompson, Noah Harding Professor of Statistics of the Department of Statistics at Rice University. “We have always been data-oriented. We deal with the real world.”
Thompson, who will retire on July 1 after 46 years, was a statistician at Rice before there was a statistics department. When he joined the faculty in 1970, after three years spent teaching mathematics at Indiana University and three years at Vanderbilt, it was as a member of the Mathematical Sciences Department in the School of Natural Sciences.
Only in 1987 did statistics become a discrete department, within the School of Social Sciences, with Thompson as the founding chair. It moved to the George R. Brown School of Engineering in 1990.
“Jim was part of the great recruiting class in mathematical sciences that saw the arrival of Richard Tapia and Ken Kennedy. He organized a statistics steering committee that recommended creation of a stand-alone department,” said David Scott, the Noah Harding Professor of Statistics and a student of Thompson’s when working for his master’s degree and Ph.D. at Rice.
Thompson earned a B.S. in chemical engineering from Vanderbilt University in 1960, and his M.A. and Ph.D.in Mathematics from Princeton University in 1963 and 1965, respectively. At Rice, his research focused on statistical model building, biomathematics, quality control and computational finance. He did pioneering work in HIV/AIDS and cancer modeling.
“Jim was one of the early Rice faculty members to become heavily involved in joint research in the Texas Medical Center, especially with biostatisticians at M.D. Anderson,” Scott said.
In 1971, Thompson began working with NASA to improve its program for using satellite data to predict Soviet agricultural production. His early work lowered the misclassification rate from 25 percent to 5 percent. Later, his former doctoral student David Scott and NASA team leader Richard Heydorn developed a four-channel system that forecast food production more accurately than the Ministry of Agriculture in Moscow.
Working with M.D. Anderson researchers H.D. Suit and Barry Brown, Thompson developed a protocol for using external-beam radiotherapy to emulate the effects of implant-radium therapy. Suit left M.D. Anderson to become the head of radiotherapy at Harvard, where the protocol became standard.
Working for the Army Research Office, Thompson built a large-scale computer model for organizing a fortified defense against Soviet attacks through the Fulda Gap in Germany. Twice he was invited to give the annual short course in applied mathematics by the Army Research Office, on the topics of empirical model building and statistical process control.
Thompson developed the SIMEST algorithm for creating multiple replicates of computer-generated pseudo-realities, which is used to estimate the parameters of the underlying model. With his student Marc Elliot he has also developed the MaxMean algorithm that permits the finding of the underlying structures of high-dimensional data sets.
In 2012 Thompson obtained a patent on a computationally intensive algorithm for portfolio optimization called the Simugram. In collaboration with Scott Baggett and John Dobelman, he developed the “Max-Median Rule for Portfolio Selection,” and continues to work on portfolio strategies with a new faculty member at Rice, Philip Ernst, assistant professor of statistics.
Thompson is a Fellow of the American Statistical Association, the Institute of Mathematical Statistics and the International Statistical Institute. He is the recipient of the Wilks Medal for his work with the U.S. Army and the Owen Award for his work in quality control.
He has directed 17 doctoral students, and authored or co-authored 14 books, including Models for Investors in Real World Markets (2002) and Empirical Model Building: Data, Models, and Reality (2nd ed., 2011). His wife, Ewa M. Thompson, is a research professor of Slavic Studies at Rice whose books have been translated into Russian, Polish, Belorussian, Ukranian, Hungarian and Chinese.
“As a young colleague of mine observed recently, Jim is always generous with his advice, and you would be crazy not to listen and follow,” Scott said.
On April 25, as a finale to the 6th Eubank Conference on Real World Markets, a banquet will be held in Duncan Hall in honor of Thompson’s retirement. Sponsored by the Center for Computational Finance and Economic Systems and the Department of Statistics, the conference is funded through an endowment by Nancy and Tom Eubank. Register here: http://www.cofes-rice.org/eubank-conferences/
Friends and colleagues have started an endowed fund for the James R. Thompson Distinguished Lectureship in Statistics. For information or to contribute to the annual lecture series, contact Christina Miller at 713-348-4612 or firstname.lastname@example.org
John A. Dobelman, professor in the practice of statistics, is assembling a Festschrift in Thompson’s honor, scheduled for publication later this year.