ENBIS: European Network for Business and Industrial Statistics

My abstracts

The following abstracts have been accepted for this event:

• Definitive Augmentation of Definitive Screening Designs
  
  Authors: Phil Kay (SAS)
  Primary area of focus / application: Design and analysis of experiments
  Keywords: Design of Experiments, Screening, Augment, Response Surface, Definitive Screening Design
  

  Submitted at 5-Feb-2018 13:42 by Phil Kay
  Accepted
  
  Definitive screening designs (DSDs) uniquely address the key needs of many experimenters. How else can we explain the rapid and enthusiastic adoption of DSDs since their discovery was published in 2011? For many experimenters, 13- or 17-run DSDs for five to seven factors are go-to designs when screening for the few driving factors. Along with design-based model selection, you potentially have a simple, efficient and effective experimental workflow to find the important main effects, interactions and curvilinear behaviors of these factors. If only three of the factors are active, you can fit the full second-order RSM model and achieve screening and optimization in one step. But what if more than three factors are active? When ambiguity occurs is there a simple next step? Or does the complexity of this situation become a barrier to adoption of DSDs? In this presentation, you will see simple ways to augment these DSDs, ensuring that the structure and properties can be preserved to maintain the benefits of DSDs. Consequently, more people in more situations can benefit from the workflow of sequential DOE and DSDs.
• Statistical monitoring of nonlinear profiles using mixed effects models
  
  Authors: Arda Vanli (Florida State University, Department of Industrial and Manufacturing Engineering)
  Primary area of focus / application: Process
  Keywords: Profile monitoring,, statistical process control,, mixed effect models,, functional data
  

  Submitted at 15-Feb-2018 23:44 by Arda Vanli
  Accepted
  

  5-Jun-2018 09:25 Statistical monitoring of nonlinear profiles using mixed effects models
  In this research we propose a profile monitoring method for sequential monitoring of dose-response curves. Dose-response curve experiments are commonly utilized in agricultural and food production applications. In most of the profile monitoring approaches, it is assumed that the errors are independent and identically distributed random variables. In many practical profile monitoring problems, however, the errors are autocorrelated. In this research we study the effectiveness of mixed-effects models to account for autocorrelation within profiles. We consider a Phase II study and investigate the effectiveness of both parametric and nonparametric regression models in monitoring nonlinear profiles. The application of the method is illustrated on a replicated dose-response experiment.
• Optimal target allocation for hypothesis testing in multiarm clinical trials
  
  Authors: Marco Novelli (University of Bologna)
  Primary area of focus / application: Design and analysis of experiments
  Keywords: Asymptotic inference;, Ethics;, Power;, Response-adaptive designs;, Wald test
  

  Submitted at 24-Feb-2018 17:22 by Marco Novelli
  Accepted
  

  5-Jun-2018 17:20 Optimal target allocation for hypothesis testing in multiarm clinical trials
  The large majority of randomized clinical trials for treatment comparisons have been designed in order to achieve balanced allocation among the treatment groups, with the aim of maximizing inferential precision in the estimation of the treatment effects. The main justification concerns the so-called ``universal optimality" of the balanced design (see e.g. Silvey, 1980), especially in the context of the linear homoscedastic model, since it optimizes the usual design criteria for the estimation of the treatment contrasts, (like the well-known D-optimality minimizing the volume of the confidence ellipsoid of the contrasts), and it is nearly optimal under several optimality criteria, also under heteroscedasticity (Begg and Kalish, 1984).
  Taking into account the problem of testing statistical hypothesis about the equality of the treatment effects, balance is still optimal in the case of two treatments, since it maximizes the power of the test for normal homoscedastic responses and it is asymptotically optimal in the case of binary outcomes (see e.g. Azriel, Mandel and Rinott, 2012; Baldi Antognini 2008). However, in the case of several treatments the balanced allocation may not be efficient, since it is significantly different from the optimal design for hypothesis testing and could be strongly inappropriate for phase III-trials, in which the ethical demand of individual care often induces to skew the allocations to more efficacious (or less toxic) treatments. To derive a suitable compromise between these goals, Baldi Antognini, Novelli and Zagoraiou suggested a constrained optimal target that maximizes the power of the classical Wald test of homogeneity, subject to an ethical constraint on the allocation proportions reflecting the efficacy of the treatments. The aim of the present work is to push forward these results, by providing some important properties of this constrained optimal allocation. The comparisons with some targets proposed in the literature show that the constrained optimal allocation has good performance in terms of statistical power, estimation precision and ethical demands and thus it represents a valid compromise between inference and ethical concerns.
  
  References
  Azriel, D., Mandel, M., Rinott, Y. (2012). Optimal allocation to maximize power of two-sample tests for binary response. Biometrika 99, 101-113
  Baldi Antognini, A. (2008). A theoretical analysis of the power of biased coin designs. Journal of Statistical Planning and Inference 138, 1792-1798
  Baldi Antognini, A., Novelli, M., Zagoraiou, M.: Optimal designs for testing hypothesis in multiarm clinical trials. Submitted.
  Begg, C.B., Kalish, L.A. (1984). Treatment Allocation for Nonlinear Models in Clinical Trials: The Logistic Model. Biometrics 40, 409--420
  Silvey, S.D. (1980). Optimal Designs. Chapman \& Hall, London
• Exploring the Latent Variable Space of a Multiresponse DOE to Optimize Solid Phase Microextraction (SPME): Case study - Quantification of Volatile Fatty Acids in Wines
  
  Authors: Marco P. Seabra dos Reis (Department of Chemical Engineering, University of Coimbra), Ana Cristina Pereira (CIEPQPF, Department of Chemical Engineering, University of Coimbra), João M. Leça (Faculty of Exact Sciences and Engineering, University of Madeira), Pedro M. Rodrigues (Faculty of Exact Sciences and Engineering, University of Madeira), José C. Marques (Faculty of Exact Sciences and Engineering, University of Madeira)
  Primary area of focus / application: Design and analysis of experiments
  Secondary area of focus / application: Process
  Keywords: Design of Experiments;, Definitive Screening Designs;, Latent variable modelling;, Principal Component Analysis;, Analytical Instrumentation;, Wine
  

  Submitted at 25-Feb-2018 21:47 by Marco P. Seabra dos Reis
  Accepted
  

  5-Jun-2018 10:15 Exploring the Latent Variable Space of a Multiresponse DOE to Optimize Solid Phase Microextraction (SPME): Case study - Quantification of Volatile Fatty Acids in Wines
  In this work we apply Definitive Screening Designs (DSD) to optimize HS-SPME extraction in order to analyze volatile fatty acids (VFA) present in wine samples. The purpose is the simultaneous optimization of seven extraction parameters (responses – the chromatographic area of nine analytes). Both qualitative and quantitative extraction parameters (factors) were considered: a two-level qualitative variable, the fiber coating, and six quantitative variables, namely the pre-incubation time, the extraction time and temperature, the headspace/sample volume, the effect of agitation during extraction and the influence of the ethanol content. Optimization w.r.t. analytes’ chromatographic responses was carried out both individually (response by response) and altogether, by modelling the responses in the latent variable space (i.e., explicitly considering their underlying correlation structure). The analysis in the latent variable space was particularly informative, allowing the simultaneous optimization of the levels of the responses and also the robustness by which the best compromise is achieved. In the end, a consensus analysis of the two perspectives was considered in the definition of the overall optimal extraction conditions for the quantification of VFA in fortified wines. The solution found was to use a DVB/Car/PDMS fiber, 10 mL of samples in 20 mL vial, 40 min of extraction at 40 C. The analysis also revealed that the factors incubation time, agitation and sample dilution do not play a significant role in explaining the variability of extraction parameters. Therefore, they were set to the most convenient levels. This work demonstrates the benefits of using DOE and latent variable modelling for the optimization of analytical techniques, contributing to the implementation of rigorous, systematic and more efficient optimization protocols.
• Choice models with mixtures: an application to a cocktail experiment
  
  Authors: Hajar Hamidouche (KU Leuven), Peter Goos (KU Leuven)
  Primary area of focus / application: Design and analysis of experiments
  Secondary area of focus / application: Modelling
  Keywords: multinomial logit model, mixed logit model, latent class model, Scheffé model, forces, hierarchical Bayes estimates, Firth estimates
  

  Submitted at 27-Feb-2018 14:40 by Hajar Hamidouche
  Accepted
  

  5-Jun-2018 11:30 Choice models with mixtures: an application to a cocktail experiment
  Choice experiments are frequently used to perform systematic studies of consumer preferences. In these experiments, the properties of products are systematically varied according to an experimental design. After the experiment has been done, the data are analyzed using choice models. Proper modeling allows a perfect understanding of the consumer preferences, the identification of consumer segments and the optimization of
  products, and, eventually, offers opportunities for competitive advantage. The literature about choice models is very extensive. However, in the analysis of choice data, products that are mixtures of ingredients have been largely overlooked. This is surprising as a great share of products, such as shampoos, cakes and cocktails, actually are mixtures. The literature on the modeling of data from traditional mixture experiments is also substantial. In this paper, we combine the theory about choice models and traditional mixture models. We will apply the resulting model to data from a real-life
  experiment in which consumers made pair-wise comparisons between seven cocktails. More specifically, we will incorporate the Scheffé model, one of the most commonly used mixture models, in three choice models. For the choice models, we will first assume consumer homogeneity. Next, we will allow for heterogeneity among individuals. Therefore, we will discuss the multinomial logit model, the mixed logit model and the latent class model. For identifying segments, besides the latent class model, we will explore a two-stage approach in which subject gradients, Hierarchical Bayes estimates and Firth individual-level estimates are used as input for a cluster analysis.
• Ban of castration and "boar taint" – does it matter?
  
  Authors: Froydis Bjerke (Animalia Meat and Poultry Research Centre)
  Primary area of focus / application: Design and analysis of experiments
  Secondary area of focus / application: Other: Food and agriculture
  Keywords: boar taint, castration, meat quality, FF-Manova, Multivariate Methods, Explorative data analysis
  

  Submitted at 28-Feb-2018 15:51 by Froydis Bjerke
  Accepted
  

  6-Jun-2018 09:30 Ban of castration and "boar taint" – does it matter?
  The EU has claimed a ban on castration of male slaughter pigs from 2018. However, the alternatives to castration are not well implemented. In order to avoid boar taint in pork meat and products thereof, the ban is postponed. What are the alternatives, and how do they compare to castration? What can be measured and what matters – for whom?
  Research on boar taint and alternatives to surgical castration has been a large issue in European animal and food research for around twenty years, increasing after the pronounced ban. Alternatives include breeding to reduce boar taint in entire males, online sorting of tainted carcasses on the slaughter line, and immunocastration (injections). Since 2016, Animalia has managed a Norwegian research project "Boars to the market – solutions for Production, Pork quality & Markers for boar taint". The main objective of this project is to initiate economically sustainable alternatives to surgical castration of male pigs. In order to achieve the main objective the following sub-goals are specified:
  1. Develop a new on-line measurement method of boar taint.
  2. Characterise quality and yield of immunologically castrated pig carcasses.
  3. Characterise consumer attitudes to and acceptance of meat from immunologically castrated pigs.
  4. Investigate DNA variation for boar taint and implement genomic breeding values.
  
  The presentation considers sub-goal 2 and how experiments and statistics are applied to investigate and quantify differences between the groups of slaughter pigs, mainly from the viewpoint of the meat industry. The analyses are quite explorative, supported by multivariate methods like Partial Least Squares and 50-50 Manova, revealing some of the challenges of measuring boar taint, and carcass and meat quality. Which traits separate the groups, and which do not?
  
  Some references
  De Briyne et al. Porcine Health Management (2016) 2:29 DOI 10.1186/s40813-016-0046-x
  Langsrud, Ø. (2002), 50-50 Multivariate Analysis of Variance for Collinear Responses, The Statistician, 51, 305-317.
  IPEMA, Innovative Approaches for Pork Production with Entire Males - a COST action (CA 15215) supported by the European Union. http://www.ca-ipema.eu/
  
  The project is financially supported from the "BIONÆR" programme of the Research Council of Norway.