Abstract:Summary
An account is given of the development of statistical computing at Rothamsted. It is concerned mainly with the period from 1954 (when the first electronic computer was delivered) until 1985 (when this article was written). Initially, many specialised programs were written, but it was soon realised that, for efficiency, general‐purpose programs—each unifying many statistical techniques—were required. The development of these programs was gradual and required corresponding developments in statistical the… Show more
“…All the preprocessing operations were performed on R [ 32 ] using in-house scripts and packages stats [ 32 ], ChemoSpec [ 33 ], and rLims [ 34 ].…”
Section: Methodsmentioning
confidence: 99%
“…All the analyses were performed using in-house scripts written for the R software [ 32 ] and based on the pROC [ 37 ] and rLims [ 34 ] packages. All source code used is available on Github ( https://github.com/jwist/coffee-profiler/ ) and as Supplementary Materials.…”
The sensorial properties of Colombian coffee are renowned worldwide, which is reflected in its market value. This raises the threat of fraud by adulteration using coffee grains from other countries, thus creating a demand for robust and cost-effective methods for the determination of geographical origin of coffee samples. Spectroscopic techniques such as Nuclear Magnetic Resonance (NMR), near infrared (NIR), and mid-infrared (mIR) have arisen as strong candidates for the task. Although a body of work exists that reports on their individual performances, a faithful comparison has not been established yet. We evaluated the performance of 1H-NMR, Attenuated Total Reflectance mIR (ATR-mIR), and NIR applied to fraud detection in Colombian coffee. For each technique, we built classification models for discrimination by species (C. arabica versus C. canephora (or robusta)) and by origin (Colombia versus other C. arabica) using a common set of coffee samples. All techniques successfully discriminated samples by species, as expected. Regarding origin determination, ATR-mIR and 1H-NMR showed comparable capacity to discriminate Colombian coffee samples, while NIR fell short by comparison. In conclusion, ATR-mIR, a less common technique in the field of coffee adulteration and fraud detection, emerges as a strong candidate, faster and with lower cost compared to 1H-NMR and more discriminating compared to NIR.
“…All the preprocessing operations were performed on R [ 32 ] using in-house scripts and packages stats [ 32 ], ChemoSpec [ 33 ], and rLims [ 34 ].…”
Section: Methodsmentioning
confidence: 99%
“…All the analyses were performed using in-house scripts written for the R software [ 32 ] and based on the pROC [ 37 ] and rLims [ 34 ] packages. All source code used is available on Github ( https://github.com/jwist/coffee-profiler/ ) and as Supplementary Materials.…”
The sensorial properties of Colombian coffee are renowned worldwide, which is reflected in its market value. This raises the threat of fraud by adulteration using coffee grains from other countries, thus creating a demand for robust and cost-effective methods for the determination of geographical origin of coffee samples. Spectroscopic techniques such as Nuclear Magnetic Resonance (NMR), near infrared (NIR), and mid-infrared (mIR) have arisen as strong candidates for the task. Although a body of work exists that reports on their individual performances, a faithful comparison has not been established yet. We evaluated the performance of 1H-NMR, Attenuated Total Reflectance mIR (ATR-mIR), and NIR applied to fraud detection in Colombian coffee. For each technique, we built classification models for discrimination by species (C. arabica versus C. canephora (or robusta)) and by origin (Colombia versus other C. arabica) using a common set of coffee samples. All techniques successfully discriminated samples by species, as expected. Regarding origin determination, ATR-mIR and 1H-NMR showed comparable capacity to discriminate Colombian coffee samples, while NIR fell short by comparison. In conclusion, ATR-mIR, a less common technique in the field of coffee adulteration and fraud detection, emerges as a strong candidate, faster and with lower cost compared to 1H-NMR and more discriminating compared to NIR.
“…Access to computing facilities was not plentiful; however, in April 1954 an Elliot 401 was installed at Rothamsted (Yates & Rees 1958). John Gower provides a useful account of programming at Rothamsted in the late 1950s and early 1960s, showing that there was quite a lot of activity (Gower 2015); John Nelder was not involved in this. In 1963, however, an Orion machine (first created by Ferranti in 1961) was acquired by Rothamsted; John started using it and he soon became an important figure in statistical computing.…”
John Nelder was a statistician who made important contributions to applied statistical theory, of which three are notable: his work on general balance; his influential collaboration with Robert Wedderburn and later Peter McCullagh on the development of generalized linear models (GLMs); and his post-retirement collaboration with Youngjo Lee on hierarchical generalized linear models (HGLMs). John started his career at the National Vegetable Research Station in Wellesbourne in 1950, and was subsequently head of statistics at Rothamsted Experimental Station, Harpenden, from 1968 to his retirement in 1984, during which time he regularly gave a course of lectures on statistical computing at Imperial College London. He wrote several highly influential books and received many honours, including the Guy Medals in Silver and in Gold of the Royal Statistical Society (in 1977 and 2005) and an honorary doctorate from the Université Paul Sabatier, Toulouse, in 1981. His work on GLMs completely changed the world of statistical modelling, his theory of general balance was years ahead of its time and it is perhaps too early to judge his work on HGLMs. The simplex algorithm for function optimization developed with Roger Mead has proved extremely popular. John was an excellent pianist with a wide range of musical interests.
“…Nelder's contributions to statistical computing at Rothamsted were described at length by Roger Payne and John Gower, who both gave (Payne, pp. 110‐111; Gower, pp. 366‐367) brief descriptions of GLIM development.…”
Section: Background To the Glm Papermentioning
confidence: 99%
“…Gavin Ross was moved to his Rothamsted slot. Gower ( p. 367) commented In 1985, because of lack of resources, work at Rothamsted associated with the GLIM project was reduced to a residual level.…”
Summary
This article documents the history of the GLIM statistical package, from Royal Statistical Society documents, GLIM Working Party documents and recollections of the developers of GLIM and its supporting papers, manuals and books.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
