ENBIS-15 in Prague

6 – 10 September 2015; Prague, Czech Republic Abstract submission: 1 February – 3 July 2015

My abstracts


The following abstracts have been accepted for this event:

  • Alternative Statistical Analysis of Interlaboratory Comparison Measurement Results

    Authors: Rossella Berni (Department of Statistics, Computer Science and Applications "G.Parenti", University of Florence), Carlo Carobbi (Department of Information Engineering, University of Florence)
    Primary area of focus / application: Other: ENBIS session at IMEKO
    Secondary area of focus / application: Metrology & measurement systems analysis
    Keywords: Proficiency test, Interlaboratory comparisons, Key-comparisons, Error measurement model
    Submitted at 29-Apr-2015 20:42 by Rossella Berni
    9-Sep-2015 09:20 Alternative Statistical Analysis of Interlaboratory Comparison Measurement Results
    Measurement results provided by laboratories involved in interlaboratory comparisons are processed in order to obtain the reference value of the quantity under measurement and its uncertainty. Measurement results are also compared with the reference value and its uncertainty in order to assess participants performance. Standards procedures are available in order to statistically process the measured values resulting from interlaboratory comparisons, such as those described in [1]. Unfortunately these procedures do not take into account the measurement uncertainty that each participant is required to declare. Well established statistical methods, such as errors measurement models, are however available that permit to statistically process both the measured values and their uncertainty, and therefore to fully evaluate the dispersion effects involved in the interlaboratory comparison activity.

    [1] Statistical Methods for Use in Proficiency Testing by Interlaboratory Comparison, ISO 13528: 2005.
  • Predicting the Evolution of Worldwide Financial Markets with the Popularity of Key Words in the Google Browser through a Compositional Approach

    Authors: Robert Ortells (Universitat Politècnica de Catalunya), Juan Jose Egozcue (Universitat Politècnica de Catalunya), Maribel Ortego (Universitat Politècnica de Catalunya), Alvar Garola (Universitat Politècnica de Catalunya)
    Primary area of focus / application: Economics
    Secondary area of focus / application: Other: Compositional Analysis of Data
    Keywords: Compositional data, Worldwide financial markets, Stock market, Sovereign debt yield, Commodities, Multiple linear regression, Google
    Submitted at 29-Apr-2015 22:17 by Robert Ortells
    7-Sep-2015 12:30 Predicting the Evolution of Worldwide Financial Markets with the Popularity of Key Words in the Google Browser through a Compositional Approach
    The present project has studied the correlation between the weekly evolution of several financial indexes from all around the globe and the popularity of several terms that were searched online in the Google, Inc. browser the previous week. Indeed, the idea that online search patterns can be explicative of immediate future performance of different economic indicators seems intuitive. For instance, it seems reasonable that in case words like crisis, bankruptcy or recession increase in popularity, then several economic indexes perform in a certain way, such as declines in the stock markets or increases in sovereign bond indexes.

    The key concept of the analysis is that a conventional multiple regression model cannot be proposed straight away, since both the explanatory and response variables have a specific nature that must be taken into consideration. With reference to the response variable (worldwide financial indexes involving equities, sovereign debt yields and commodities), we can see that the important information is not in the actual values for the indexes but in the relationships between them. Indeed, performance of financial indexes has to be evaluated through comparing them with respect to other indexes. For example, a positive evolution for a stock market index needs to be analyzed by looking at other stock market indexes (apart from other relevant financial indexes), otherwise erroneous conclusions might be drawn. Therefore, since the relevant information is not in the numerical values but in the relationships with the rest, a compositional approach seems appropriate. Regarding the information of the popularity of the words in the internet browser, the same reasoning seems to be adequate. For instance, in case the word bankruptcy increases in popularity for a certain week and the other words present a similar increase, the information we might infer from such search patterns is the same. However, in case the searches increase for such word and remain quite the same for the rest, this may be relevant. Therefore, the information on the popularity of the different terms presents a compositional nature as well, so it must be treated accordingly before performing any regression.

    For the reasons we have just set, a compositional approach seems necessary in order to address the problem successfully, since both the explanatory and the response variables present a compositional nature. Once the data has been appropriately treated, an exploratory analysis has been performed on both data sets (popularity of 200 different words and numerical values of 19 financial indexes along the period 2004-2014). After that, a multiple linear regression model between them has been proposed. Finally, the results have been compared to previous research in the field.
  • How Well Do We Need to Measure Our Drilling Mud?

    Authors: Winfried Theis (Shell Global Solutions International B.V.)
    Primary area of focus / application: Design and analysis of experiments
    Secondary area of focus / application: Quality
    Keywords: Measurement uncertainty, Design of Experiments, Computer experiments, Quality by design
    Submitted at 30-Apr-2015 03:06 by Winfried Theis
    8-Sep-2015 17:20 How Well Do We Need to Measure Our Drilling Mud?
    Shell is always working to improve the quality of well construction operation, to ensure safe exploration of and production from the gas and oil reservoirs around the world. At the same time it is necessary to reduce costs as it becomes necessary to drill more wells to access smaller or unconventional reservoirs in an economical manner. One important part of the drilling operation is the hydraulic system formed by the drilling mud that is used to transport the cuttings from the hole. In this project the goal is to develop an online measurement system for the drilling mud. When considering the choice of the sensors the question arose, what is the measurement system performance requirement for the sensors on the hydraulic system for sufficient control of the outputs. As the hydraulic system is a complex non-linear system a simulator based on first principles was used to learn how random disturbances are propagated through the system. A series of designs of experiments for computer simulations was used to determine the most important influences on the sensitivity of the hydraulic system and subsequently to find safe upper limits for the measurement uncertainties.
  • Introduction to Compositional Data Analysis with Applications to Customer Survey Analysis

    Authors: Marina Vives-Mestres (Universitat de Girona), Josep-Antoni Martín-Fernández (Universitat de Girona), Ron Kenett (KPA Group)
    Primary area of focus / application: Other: Compositional Analysis of Data
    Keywords: Customer survey analysis, Compositional data analysis, Logratio coordinates, Simplex
    Submitted at 30-Apr-2015 07:28 by Marina Vives-Mestres
    7-Sep-2015 11:30 Introduction to Compositional Data Analysis with Applications to Customer Survey Analysis
    Compositional Data (CoDa) consists of multivariate data with strictly positive components of constant sum. Natural examples consist of chemical formulations or results from a survey. We first introduce the definition of CoDa through simple examples and then show the need of an analysis based on log ratios of components, also called coordinates. We then describe the principles of working with coordinates and graphically compare classical analysis with a log ratio analysis in the cases of: principal component analysis, cluster analysis, linear discriminant analysis and linear regression models. We also discuss the problem of zeros and how to deal with them in CoDa.

    We finish this introductory session with an example of applications in the field of customer survey analysis. Specifically, we analyse the annual customer satisfaction survey of the ABC Company presented and analysed in detail in the book edited by Kenett and Salini (2011). The questionnaire consists of an assessment of overall satisfaction evaluated on a five-point anchored scale, so that it can be analysed from a CoDa perspective, and almost 50 statements with two types of scores: an evaluation score and a measure of item importance. Other questions such as repurchasing intentions and descriptive variables for each customer are used in analysing the ABC dataset.

    We show how CoDa methods can contribute to provide a map of customer’s opinion, improve decision making, identify improvement areas or weak points, set service level targets and help improve the questionnaire itself. We also compare the findings to several statistical models presented in Kenett and Salini (2011) such as PLS, hierarchical models, fuzzy sets, log-linear models and control charts. Graphical tools to communicate CoDa results are also proposed. The general idea is that one can increase the information quality of a customer survey analysis by combining more than one technique.

    Following this introductory talk, a special CoDa session will include two additional application examples as well as a practical and interactive CoDa hands-on session. The audience is highly encouraged to attend the special CoDa session and enjoy the hands-on experience that will be delivered.


    Kenett, RS., Salini, S. (2011). Modern Analysis of Customer Satisfaction Surveys: with applications using R. Chichester: UK. JohnWiley and Sons.
  • On- and Offline Detection of Structural Breaks in Thermal Spraying Processes

    Authors: Nikolaus Rudak (Dortmund University of Applied Sciences and Arts), Matthias Borowski (Institute of Biostatistics and Clinical Research), Birger Hussong (Faculty of Mechanical Engineering), Dominik Wied (TU Dortmund University), Sonja Kuhnt (Dortmund University of Applied Sciences and Arts), Wolfgang Tillmann (Faculty of Mechanical Engineering)
    Primary area of focus / application: Process
    Keywords: Jumps, Trends, Variance changes, Thermal spraying process
    Submitted at 30-Apr-2015 09:06 by Nikolaus Rudak
    7-Sep-2015 17:00 On- and Offline Detection of Structural Breaks in Thermal Spraying Processes
    Industrial processes, like for example thermal spraying processes, are often supervised by several variables that can be measured online. In the case of a thermal spraying process, particle properties are recorded during the experiment. These time series contain valuable information about the coating quality (Rudak et al. (2012)). Technical malfunctions typically lead to structural breaks like jumps and trend changes in the time series. They can also induce changes in the variance of the time series. The detection of such structural breaks helps the engineer to react so that production failures can be avoided.

    We use a robust online filtering procedure for detecting jumps in the mean and modify this method slightly in order to detect ongoing trends. Furthermore, we utilize a fluctuation test for constant variances. We investigate the mentioned methods by simulations and apply them to data coming from experiments where the engineer provokes several technical malfunctions during the thermal spraying process (Borowski et al. (2014)).


    1. Rudak, N., Kuhnt, S., Hussong, B. and Tillmann, W. (2012), "On
    different strategies for the prediction of coating properties in a HVOF process", SFB 823 Discussion Paper 29/12, TU Dortmund University.

    2. Borowski, M., Rudak, N., Hussong, B., Wied, D., Kuhnt, S., Tillmann, W. (2014), "On- and offline detection of structural breaks in thermal spraying processes", Journal of Applied Statistics, 41 (5), 1073-1090.
  • A Discussion on the Concept and Objectives of Blocking and Robustness in Industrial Experiments

    Authors: Xavier Tort-Martorell (UPC Universitat Politécnica de Catalunya, Barcelona Tech), Lluís Marco-Almagro (UPC Universitat Politécnica de Catalunya), Pere Grima (UPC Universitat Politécnica de Catalunya)
    Primary area of focus / application: Design and analysis of experiments
    Keywords: Industrial experiments, Factorial designs, Blocking, Robustness, Taguchi
    Submitted at 30-Apr-2015 09:48 by Xavier Tort-Martorell
    Accepted (view paper)
    7-Sep-2015 10:20 A Discussion on the Concept and Objectives of Blocking and Robustness in Industrial Experiments
    The need and the advantages of blocking experimental designs have been well justified and documented since Fisher introduced the concept. The most influential industrial statistics books (Box et al., Montgomery, Wu et al., Kenett et al.) remark their importance to reduce or eliminate the effect of suspected or known sources of undesirable changes in the homogeneity of experimental conditions. Those books also remark the importance of robust systems, defined as insensitive to the varying conditions of its use in the real world.
    In this presentation we will argue that in many real world situations, and in fact in all the examples used in the above mentioned books to illustrate blocking, it is a better conceptual and practical option to design an experiment with the aim of achieving robustness than to block the design.
    Attention will be devoted to the rather different philosophy/intuition underlying the two approaches. In the case of blocking it is assumed that the blocking factors do not interact with the other experimental factors, while in the case of robustness the hope is that the noise factors interact with the control factors.