Seminar Schedule: Year 2021

  •  Friday, feburary the 26th  2021. Time: 12:00

Juan José Egozcue. Emeritus Professor at the UPC

From the Nightingale chart to compositional biplots





From the Nightingale chart to compositional biplots

SPEAKER: Juan José Egozcue


PLACE Online. Register by February 24 at the latest.

 DATEFriday, february the 26th, 2021. Time: 12:00

ABSTRACT:  Among the remarkable merits of Florence Nightingale (1820--1910) the introduction of her chart presenting casualty data of Crimea War (1854-1856) is notorious. Her intention was to show to politicians and public in general that most deaths were avoidable taking measures of hygiene and care, in front of deaths caused by war wounds or other causes. Remarkably, the chart---that she called coxcomb and is nowadays known as polar area diagram---takes into account two important points in any statistical research: The research question and the characteristics of the available data. The data presentation on a polar plot accounts for seasonality; although major attention is paid to avoidable deaths, they are always compared to other causes of death, thus inviting to a multivariate analysis, where both proportions of types of deaths and their total are considered. This suggests a discussion on the sample space of the data. Available log-ratio approaches for the analysis of compositional data allow an alternative presentation emphasising the ratios between the different components, but not forgetting the total number of casualties. The compositional biplot including the total of the components is, maybe, the richest representation in the context of compositional data analysis.


 Juan Jose Egozcue: He studied Physics, at the University of Barcelona(Spain). He obtained his PhD in 1982. In 1978 he got lecturer in the Civil Engineering School (Universitat Politècnica de Catalunya, UPC, Barcelona, Spain), teaching subjects on Applied Mathematics and Statistics. Full Professor in 1989, at the UPC. Present main research activity: Bayesian methods for natural hazard assessment; and analysis of compositional data, with special emphasis in the geometry of the sample space. After retirement in 2016, he is Emeritus Professor at the UPC and President of the CoDa-Association.

Vera Pawlowsky-Glahn: She studied Mathematics at the University of Barcelona (Spain) and received her PhD from the Free University Berlin (Germany) in 1986. Back to Spain, she was professor at UPC in Barcelona and later professor of Statistics with the University of Girona. Since 1982 her main research field is the statistical analysis of compositional data, with special emphasis in the algebraic geometric structure of the sample space. She has published numerous articles, proceedings and books on this topic. She was President of IAMG 2008-2012 and President of the CoDa-Association 2015-2017. She is since November 1, 2018 Emeritus Professor at the University of Girona, Spain.