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:

  • Teaching Practical Forecast Methods: Learning by Doing; a Twenty Years Experience

    Authors: Annie MORIN (Université Rennes-1) Jean-Hugues CHAUCHAT (Université Lyon-2)
    Affiliation: Universités de Rennes et de Lyon, FRANCE
    Primary area of focus / application: Education & Thinking
    Keywords: Teaching , Microeconomics , Forecasting , Consulting , Communication
    Submitted at 18-Apr-2011 17:02 by Jean-Hugues Chauchat
    Accepted (view paper)
    7-Sep-2011 12:00 Teaching Practical Forecast Methods: Learning by Doing; a Twenty Years Experience
    A difficult part of teaching statistics is to push students to make the link between theoretical models and real world situation. A large part of our efforts, at graduate and post-graduate level of statistical education, is to put students in the following situation: they have to understand what a non-statistician client is asking for, to meet client's needs with realistic solutions.
    During our lectures of “microeconomics forecasting methods”, our students, in groups of two, have to find a shop, a company, a community organisation, etc., who gives them time series of sales or activities in order to make models and calculate forecast for the future days, weeks or months.
    The best situation is when students find their client by themselves. In this situation, students are motivated to work a lot and the teacher becomes the student's "statistical consultant".
    We shall present different case studies, the students’ difficulties and what they learn during making forecast for real non-academic clients, in Rennes and Lyon (France) and Kharkov (Ukraine) Universities.
  • Improving Quality of Annealing Process - Case Study

    Authors: Msc. Eng. Michal Rogalewicz, Ph. D. Agnieszka Kujawinska
    Affiliation: Poznan University of Technology
    Primary area of focus / application: Quality
    Keywords: quality improvement , design of experiments , dominance-based rough sets approach , annealing process
    Submitted at 20-Apr-2011 12:45 by Michal Rogalewicz
    7-Sep-2011 12:40 Improving Quality of Annealing Process - Case Study
    In the paper some aspects of project concerning annealing process were discussed. This process is carried out on forgings devoted to surgical instruments and its task is to ensure their proper hardness. Unfortunately company X has some problems with variability of this most important feature. Too high hardness causes a damage of expensive machining tools which are used to mill annealed forgings. That is why very often the operation of annealing is made twice (what of course is an additional cost for the company).
    In the project any factors having an influence on annealing process and a plan of experiment leading to specify which of them are the most important were developed. To better know the process a trial to match empirical probability distributions of main factors to known theoretical ones was made.
    Building a process knowledge is one of the most important tasks of process engineer. To cope with this challenge he needs some tools which help him. Authors try to implement Dominance-based Rough Set Approach DRSA) to this process and get some rules describing it.
    In the paper a description of this project and some main conclusions was included.
  • Identifying Sequences of Behaviour in Public Service Data

    Authors: Shirley Coleman and Matteo Bossi
    Affiliation: Newcastle University
    Primary area of focus / application: Business
    Keywords: datamining , sequences , public , service , data
    Submitted at 26-Apr-2011 01:38 by Shirley Coleman
    5-Sep-2011 16:41 Identifying sequences of behaviour in public service data
    In many areas of public service, extensive data are collected on uptake of facilities and activities, attendance at events and other measures of involvement. For example, health care providers record how many agency personnel are employed, service providers count numbers of claims, leisure centres monitor attendance. Snapshots such as these taken at a moment in time allow useful comparisons between time periods, for example in statistical process control (SPC) or between peer groups, for example in funnel plots for health care providers. Longitudinal studies include the time dimension and follow the progress of individual units which can be people, places or situations. Interest then focuses on sequences of events, such as in basket analysis where associations and sequences of purchases are analysed, or in cell kinetics where the time spent at each stage of the cell cycle is studied. In the latter case, the events within the cell cycle represent a composition with % time spent at each stage being important. In cell kinetics, the stages are fixed and it is the % time spent in each which changes. In service uptake, the % time in each activity changes, as does the number of activities and the overall time period to be considered. Unlike purchases in basket analysis, the longevity of each event is also of interest. In practical terms, as well as ongoing costs for each event, there may also be set-up costs. So, for example 3 weeks in one situation is conceptually different to 3 separate weeks in the same situation. We examine a dataset of sequences of time spent in different activities by recipients of different public services. We focus on detecting recurring sequences as well as outliers. Later work will aim to identify clusters of recipients on the basis of their sequences so that they can be provided with tailor made assistance and better customer service. The work has applications to service providers in other sectors, such as health care and finance.
  • Black-and-white Thinking Re-emerges in Quality Risk Management

    Authors: Antje Christensen
    Affiliation: Novo Nordisk
    Primary area of focus / application: Quality
    Keywords: Quality risk management , FMECA , Risk grid , Manufacturing , Continuous approach to specifications
    Submitted at 26-Apr-2011 10:06 by Antje Christensen
    Accepted (view paper)
    6-Sep-2011 15:25 Black-and-white Thinking Re-emerges in Quality Risk Management
    Quality risk management is gaining importance as a tool for quality assurance and process improvement. Most techniques, including the popular Failure Mode, Effects, and Criticality Analysis (FMECA) and the risk grid, rely on point estimates for the probability of occurrence of failure, the probability of detection before harm is done, and the severity of the consequences. However, in many cases these parameters vary for the same failure mode: Small deviations from the normal mode of operation may be rather probable, but carry little severity, while the same failure mode may lead to much larger deviations from the normal, with lower probability of occurrence, but more severe consequences. In these cases, a curve in the risk grid may be more appropriate to quantify a risk than the usual dot. The use of a point estimate in these cases is reminiscent of the black-and-white thinking refuted by early proponents of the quality movement, where any deviation from specification limits was treated equally. Statistics made important contributions to process optimisation through a continuous approach to in-spec production. Similarly, statisticians may be able to expand our understanding of risk by introducing a continuous approach to risk management. This talk will show a few examples and invite further research.
  • Re-visited: Sampling Design (not only) for Nutrition Intervention Studies

    Authors: Winfried Theis, Maria Andersson, Guus Duchateau
    Affiliation: Shell Global Solutions, ETH Zürich, Unilever R&D
    Primary area of focus / application: Design and analysis of experiments
    Keywords: Random Sparse Sampling , Optimal Design , Food Industry , Mixed Effect Modelling
    Submitted at 27-Apr-2011 13:37 by Winfried Theis
    6-Sep-2011 12:30 Re-visited: Sampling Design (not only) for Nutrition Intervention Studies
    Four years ago in Dortmund we presented a simulation study investigating the best sampling strategy to investigate the dynamic development of body stores. The major obstacle to run such a study was the high variability of the individual developments and the fact that only few samples (<=5) could be taken per subject. Optimal experimental design led to random uniform sampling, and the simulation study showed that there was no loss of the optimality due to practical restrictions.
    In the meantime we run a highly successful study on body iron stores. In this presentation we want to discuss the conclusions from this study mainly from the statistical point of view. The study design and analysis was able to convince experts in the field to consider a change in their approach to designing such studies. The most convincing argument for the nutrition experts was the fact that even very small differences in the population mean development were picked up significantly. The other major advantage of the random sampling, was the fact that visit-re-scheduling was not impairing the study design asit would have been for other design schemes.
  • Online Design of Experiments applied to an Industrial Packaging Machine

    Authors: Koen Rutten Josse De Baerdemaeker Bart De Ketelaere
    Affiliation: Laboratory of Mechatronics, Biostatistcs and Sensors, Department Biosystems, KULeuven, Kasteelpark Arenberg 30, 3001 Heverlee
    Primary area of focus / application: Design and analysis of experiments
    Keywords: DOE , optimization , EVOP , RSM
    Submitted at 27-Apr-2011 13:40 by Koen Rutten
    5-Sep-2011 16:30 Online Design of Experiments applied to an Industrial Packaging Machine
    The general goal of the research is to develop sequential Design of Experiments (DOE) methods that can be used during the production process, without having to stop the process. They differ from classical DOE because new experiments are incrementally defined based on the outcome of previous experiments, a concept that was proposed by George Box in 1957 (“Evolutionary Operation”, EVOP). This work describes the practical use of this on-line experimentation strategy for process optimization, and the common pitfalls related to it. A commercial Vertical-Form-Fill-Seal (VFFS) system is used as the practical case study. A connection is made between the newly written optimization software, which was named OpCalc, and the machine via an OPC connection so that OpCalc can adjust the settings of the machine dictated by the DOE approach. As response variable the quality of the packages was used. Results based on first experiments are presented and discussed.