Abstracts & Timetable (UTC+1)

14h00 - 14h05

    Introduction

 

14h05 - 14h35

 

  • Speaker: Alessandra Menafoglio
  • Title:  Object Oriented Spatial statistics in Bayes spaces: from distributional data to phase-amplitude variability
  • Abstract: In the presence of increasingly massive and heterogeneous spatial data, geostatistical modeling of distributional observations plays a key role. Choosing the ‘right’ embedding space for these data is of paramount importance for their statistical processing, to account for their nature and inherent constraints. The Bayes space theory is a natural embedding space for (spatial) distributional data, and was successfully applied in varied settings. In this presentation, we shall discuss the state-of-the-art methods for the analysis and prediction of georeferenced distributional data, when their spatial dependence cannot be neglected. We shall consider the viewpoint of object-oriented spatial statistics (O2S2), a system of ideas for the analysis of complex data with spatial dependence. In the broad context of O2S2, a focus will be also made on the analysis of spatial functional data (not necessarily distributional) characterized by both phase and amplitude variability, showing that the Bayes space theory provides a natural embedding for the so-called warping functions, that are used to assess the possible phase variability of the data. All the theoretical developments will be illustrated through their application on real environmental data, highlighting the intrinsic challenges of O2S2 and of the Bayes spaces approach.

 

14h35 - 15h05

 

  • Speaker: Berta Ferrer Rosell
  • Title: Compositional analysis of tourism-related content
  • Abstract: CoDa has started to be used in several fields of Social Sciences, which often face compositional research questions. For instance, in marketing or communication, typical research questions are related either to the distribution of a whole (market share, product portfolio, consumer spending distribution) or to the relative importance (e.g., advertising content or style, preferred product attributes, dominance of contents posted on different sources). In the particular field of tourism, how tourism destinations allocate resources to their product portfolios or how tourists visiting the destination distribute their tourism budget in activities with different economic impacts at the destination (undertaking different type of cultural activities, consuming at local restaurants, etc.) are questions researchers would be interested in studying. Another example are the contents posted by managers in many platforms, which may affect both consumers’ choices and satisfaction. This talk focusses on the analysis of content posted by destinations, companies, tourists, media, travel guides on several sources such as social media (e.g., Facebook), corporative websites, user opinion platforms (e.g., Trip Advisor), and also considering the type of content posted (e.g., destination image attributes, complaints, emotions), as well as, considering the type of language used. It shows how the CoDa tools (compositional distance, biplot and MANOVA) have been applied to respond to tourism researchers’ interests.

 

15h05 - 15h35

 

  • Speaker: Jennifer McKinley
  • Title:  Compositional analysis using balances of geochemical environmental toxins to explore potential associations with chronic kidney disease
  • Abstract: Digital spatial data can be used to explore potential relationships between naturally occurring geogenic elements in soil and water, potentially toxic elements (PTEs) and long term cumulative exposure which may be linked to chronic disease. However, spatial data sets such as geochemical survey data pose many challenges for exploratory data analysis.  Geochemical data are compositional in nature in that they convey relative information. As a result, compositional data analysis (CoDA) methods are frequently used to extract information from geochemical data by treating log ratio or equivalently transformed data instead of analysing the raw constant sum values. Using an urban soil geochemistry database of total element concentrations of potentially toxic elements (PTEs), and the UK Renal Registry which provided Standardised Incidence Rates (SIRs) of Chronic Kidney Disease (CKD), a statistical relationship was found between CKD of uncertain aetiology (CKDu) and environmental toxins. This study investigates the use of a compositional balance approach to determine the relative influence of geogenic and anthropogenic sources on the urban geochemistry signature.

 

 

15h35 - 16h05

 

  • Speaker: Luz Calle Rosingana
  • Title: Variable selection in microbiome compositional data analysis
  • Abstract: Though variable selection is one of the most relevant tasks in microbiome analysis, e.g. for the identification of microbial signatures, many studies still rely on methods that ignore the compositional nature of microbiome data. The applicability of compositional data analysis methods has been hampered by the availability of software and the difficulty in interpreting their results. In this talk I will focuse on three methods for variable selection that acknowledge the compositional structure of microbiome data: selbal, a forward selection approach for the identification of compositional balances, and clr-lasso and codalasso, two penalized regression models for compositional data analysis. I will discuss the link between these methods and their advantadges and limitations for variable selection in the context of microbiome studies.

 

 

16h05 - 16h35

 

  • Speaker: Željko Pedišić
  • Title:  Should compositional data analysis be used in time-use epidemiology?
  • Abstract: Time use is a health-related factor that every single person is inevitably exposed to every day, 24 hours a day. From a public health perspective, it is therefore important to investigate determinants, incidence, distribution, and effects of health-related time-use patterns in the population, and methods for preventing unhealthy time use and achieving the optimal distribution of time for population health. This presentation will include:
    [i]  a summary of the framework for Viable Integrative Research in Time-Use Epidemiology (VIRTUE);
    [ii] examples of the use of compositional data analysis (CoDA) in time-use research; and
    [iii] highlights from a recent debate between a strong proponent and a vigorous opponent of the use of CoDA.
    The presentation will conclude with recommendations on what to do to facilitate (or discourage) the use of CoDA in time-use epidemiology.

 

16h35 - 16h40

   Conclusion

From 16h40

   Chat rooms