Theo Gasser

Theo Gasser died on October 1, 2023 in Zürich, Switzerland, where he was Professor Emeritus of the Institute of Biostatistics at the University of Zürich, at the age of 82.

Theo Gasser was born on May 9, 1941, in Rüti, Switzerland. Gasser established Biostatistics at the University of Zürich starting in 1991, retiring in 2006, after serving as Professor at the University of Heidelberg, Germany, 1976-1991, where he was affiliated with both the Zentralinstitut für Seelische Gesundheit of the Medical Faculty Mannheim, and the Institute of Applied Mathematics in the Department of Mathematics.

He made major contributions to the area now known as Functional Data Analysis (FDA) and also was a main contributor to the methodology and theory of kernel methods for smoothing and differentiation (“Gasser–Müller kernel estimator”). His PhD advisor at the ETH Zürich in Switzerland was Peter Huber, the pioneer of modern robust statistics, who had begun to embrace data analysis when he became the advisor of Gasser and other students at the ETH. Gasser also made numerous contributions to statistical EEG analysis and auxology [the study of all aspects of human physical growth], the latter through a close association with the Zürich Longitudinal Growth Study. Throughout his career he was committed to exploratory data analysis with computational emphasis.

Theory for FDA was crafted in 1950 by Grenander, emphasizing Karhunen’s 1949 basis expansion in the context of Gaussian processes, which is now known as Karhunen–Loève expansion. In this paper, Grenander presented the now ubiquitous functional linear model as a consequence of the Gaussianity of stochastic processes. Another key contribution to the nascent field of FDA was the development of functional principal component analysis in Hilbert spaces by Kleffe in 1973, and a French school of FDA emerged in the early 1980s in Toulouse.

Gasser’s contributions to the early development of FDA were independent from this previous work and led to several highly cited papers. They brought entirely new dimensions to this field and were motivated by his commitment to analyze longitudinal growth curves. This led him to a framework of FDA where he not only recognized the importance of smoothing in the context of functional data but also the need to develop estimators for derivatives in order to quantify the underlying dynamics. Other topics that he pioneered in the context of FDA included robust smoothing, time warping and the utility of landmarks and shape-invariant modeling for samples of curves. These developments had a major impact on subsequent developments in FDA, where Jim Ramsay eventually coined the term “Functional Data Analysis”. Nowadays FDA has become a major field in statistics with many ongoing developments and new data challenges. Theo Gasser was one of its founders.

—
Written by Alois Kneip, University of Bonn, Germany, and Hans-Georg Müller, University of California, Davis
In the January/February 2024 IMS Bulletin issue, where this obituary was printed, Prof. Gasser’s birth year was wrongly typed as 1949 instead of 1941. We apologise for any confusion caused.