ENBIS-11 in Coimbra

4 – 8 September 2011 Abstract submission: 1 January – 25 June 2011

My abstracts


The following abstracts have been accepted for this event:

  • A logistic regression model for the risk of musculoskeletal complaints in the lumbar region

    Authors: Anabela Silva, Paula Carneiro, Ana Cristina Braga, Paulo Sampaio and Mónica Barroso
    Affiliation: Universidade do Minho
    Primary area of focus / application: Modelling
    Keywords: WRMSD (Work related musculoskeletal disorders) , logistic regression , OR (Odds ratio) , ROC (Receiver Operating Characteristic)
    Submitted at 14-Jun-2011 18:03 by Ana Cristina Braga
    5-Sep-2011 16:45 A logistic regression model for the risk of musculoskeletal complaints in the lumbar region
    Work related musculoskeletal disorders (WRMSDs) associated to repetitive work, awkward postures and other factors are currently great concern in most industrialized countries.
    In order to assess the risk of WRMSDs in nurses, we elaborated a questionnaire to gather the opinion of nurses exercising their profession in health centers in the North of Portugal, regarding musculoskeletal complaints associated with the performance of their duties.
    From a total of 3017 professionals we obtained 147 completed questionnaires that correspond to a response rate of 7.16%. Of these 147 nurses, 125 provide support at home.
    The body areas with more complaints are the back and shoulders. There is statistically significant association only between “musculoskeletal complaints in the lumbar region” and “provide home care” (OR=3.185, p-value<0.05), and a corresponding CI = [1.256, 8.075]. Based on these results we can say that nurses who provide home-based care have a chance to have musculoskeletal complaints in the lumbar region three times higher when compared with the which do not provide such support.
    To build a model associated with this complaint, we used logistic regression techniques. In this sense, it made a univariate analysis to select where the variables (factors) that may contribute to the onset of complaints in nurses who provide home-based care. We used the stepwise technique with input and output p-values, respectively, of 0.20 to 0.25, to include factors that could contribute to the model. A diagnostic analysis of the adopted model was performed using the ROC curve analysis.
  • Linear Models, Zernike Polynomials and Applications to Wafer Data

    Authors: A. Di Bucchianico B.J. Janssen
    Affiliation: Eindhoven University of Technology
    Primary area of focus / application: Modelling
    Keywords: linear regression model , circular data , Zernike polynomial , R package
    Submitted at 15-Jun-2011 12:54 by Alessandro Di Bucchianico
    Accepted (view paper)
    5-Sep-2011 10:10 Linear Models, Zernike Polynomials and Applications to Wafer Data
    The wafer technique in the semiconductor industry requires statistical analyses of data sets with a spatial aspect, since wafers have disk-like forms. We will discuss industrial case studies that show that the data may not only be scalar-valued, but also be in the forms of profiles. Because of the rotational symmetries, standard polynomial linear regression models are not appropriate. We will discuss why regression models in terms of Zernike polynomials seem to be more appropriate. An R package for fitting Zernike polynomials will be shown. We will also discuss both practical and theoretical issues in fitting regression models with Zernike polynomials.
  • Applications of Statistics in the Field of Energy

    Authors: Christian Derquenne
    Affiliation: EDF R&D
    Primary area of focus / application: Modelling
    Keywords: Statistics fields , Energy management , Times series , Segmentation , Non stationarity
    Submitted at 16-Jun-2011 09:57 by Christian Derquenne
    Accepted (view paper)
    5-Sep-2011 11:50 Applications of Statistics in the Field of Energy
    The EDF Group is one of the world’s leading energy companies, active in all areas from generation to trading and network management. It has a sound business model, evenly balanced between regulated and deregulated activities. With its first-rate human resources, R&D capability, expertise in engineering and operating generation plants and networks, as well as its energy eco-efficiency offers, the Group delivers competitive solutions that help ensure sustainable economic development and climate protection. The R&D Division aims to consolidate and develop a production mix low-carbon (nuclear advantage, development of renewable energy, capture and carbon storage), control energy demand (knowledge of the application, new uses for electricity, sustainable city ) and adjust the electrical system (asset management systems, development of transport infrastructure). Research programs are turning directly to the Group's businesses: production, energy management, business development, power systems, renewable energy and transversely by information technology. The choice of the statistical approach has been made of the establishment of R&D (1946). The three main poles and its applications are: life sciences (environmental and climate risks, and health), engineering (industrial hazards, electricity, production, consumption and pricing) and science tertiary (sales and marketing, sociology, and econometrics and financial risk). This conference will present various examples from these three main areas of applications of statistics within the EDF Group, and then we will introduce a segmentation method developed for time series analysis of market prices in the management of the energy.
  • Control of Dependence for Heavy Tailed Bivariate Data

    Authors: Zempléni, András
    Affiliation: Eötvös Loránd University, Budapest
    Primary area of focus / application: Modelling
    Keywords: control , copula , dependence , E-statistics , heavy tailed , probability integral transform , stock index data
    Submitted at 16-Jun-2011 15:56 by András Zempléni
    Accepted (view paper)
    6-Sep-2011 12:50 Control of Dependence for Heavy Tailed Bivariate Data
    We consider possible alarm procedures, based on bivariate data.
    Traditional approaches use either the mean or the covariance matrix.
    However, there are situations, where besides these features the nonlinear dependencies also play an important role, like in the case of heavy-tailed data. This
    is described best by the so-called copulae, which allow for the
    separation of the univariate modeling from the dependence. The changes in the dependence structures
    can be captured by different methodologies, including a variant of the so--called
    E-statistics, [1]. We compare this method with other approaches, for example the probability integral transform-based procedure in [2].

    The results are illustrated by stock index data.

    [1] Szekely, G. J. and Rizzo, M. L. (2005) A New Test for Multivariate Normality. Journal of Multivariate Analysis, 93/1, 58-80.
    [2] Rakonczai, P. and Zempléni, A. (2007) Copulas and goodness of fit tests. In: Recent Advances in Stochastic Modelling and Data Analysis, World Scientific Publishing, pp. 198-205.
  • The use of EFQM’ criteria on the physical activity programmes for elderly people: results of a cluster analysis

    Authors: Ana I Marques MSc1; Maria J Rosa PhD2; Marlene Amorim PhD2; Pedro Soares PhD; António Oliveira-Tavares MSc1; Jorge Mota PhD1; Joana Carvalho PhD1
    Affiliation: 1 CIAFEL - Faculty of Sports, U.Porto ; 2 Department of Economics, Management and Industrial Engineering - University of Aveiro, Portugal
    Primary area of focus / application: Quality
    Keywords: Quality improvement , EFQM , Physical activity programmes , Elderly , Cluster analysis
    Submitted at 17-Jun-2011 18:07 by Ana Marques
    Accepted (view paper)
    5-Sep-2011 10:50 The use of EFQM’ criteria on the physical activity programmes for elderly people: results of a cluster analysis
    Quality is an important issue when designing a physical activity (PA) programme for older people. The EFQM Excellence Model has been widely suggested as a framework for evaluating the quality of an organization. In this study we apply this model to the context of PA programmes for older people in Portugal. The study sets up: 1) to distinguish groups of PA programmes according to their implementation of the quality management practices (QMP) associated with the model criteria, and 2) to provide an exploratory characterization of the identified groups concerning the profile of the programmes’ general characteristics.

    A methodological triangulation was conducted in 26 PA programmes using questionnaire surveys, semi-structured interviews and document analysis. Cluster analysis using Ward's method of agglomeration with squared-Euclidean distance measures was used to identify subgroups of PA programmes based on the results of the implementation of QMP. The existence of statistically significant differences among the four subgroups in terms of programmes’ characteristics was tested resorting to the chi-square and Kruskal-Wallis tests.

    We identified four clusters of PA programmes. They differ in the degree of implementation of each EFQM’s criterion (intensity) and on the number of criteria addressed (variety). No statistically significant differences were found between clusters for their general characteristics, except for the number of facilities managed by the programme (p≤0.05).

    Clustering identified four subgroups of PA programmes based on the intensity and variety of implementation of QMP associated with the EFQM model criteria. Since the quality of a service increases customer satisfaction, the continuous improvement of the PA programmes should be addressed to increase elderly satisfaction and adherence. The identification of these clusters may help PA programmes’ leaders to learn ones from the others in order to improve the quality.
  • Modern Pharmaceutical Development and Manufacturing: A Decade into Using Multivariate Data Analysis

    Authors: Jose C. Menezes
    Affiliation: IST, Technical University Lisbon Institute for Biotechnology and Bioengineering Av Rovisco Pais, 1049-001 Lisbon, Portugal
    Primary area of focus / application: Mining
    Keywords: multivariate data analysis , multivariate modeling , pharmaceutical process development , pharmaceutical manufacturing
    Submitted at 20-Jun-2011 10:39 by Jose Cardoso de Menezes
    6-Sep-2011 10:45 Modern Pharmaceutical Development and Manufacturing: A Decade into Using Multivariate Data Analysis
    The use of multivariate data-based tools for exploring existing data, to design experiments to populate and expand the “knowledge space”, to model processes and be used in different ways throughout process development and routine manufacturing is reviewed for world-class pharmaceutical companies.

    The foundation for such efforts has known an exponential increase since the US FDA launched a comprehensive effort for opening the pharma and biotech industries to innovation and the use of best available technologies (FDA, 2004 - Guidance for Industry: Process Analytical Technology, A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance).

    In this talk based on our own work with diverse companies in EU in the past decade we will describe how multivariate techniques are transforming the landscape of process systems engineering in a previously very conservative sector. The use of diverse variable reduction and projection methods will be illustrated for unsupervised (exploratory) situations and supervised (modelling and scale up) case studies. Then a discussion into the future addressing challenges and opportunities for disciplines at the core of this revolution such as applied multivariate statistics and systems engineering will be presented.