ENBIS-13 in Ankara

15 – 19 September 2013 Abstract submission: 5 February – 5 June 2013

My abstracts


The following abstracts have been accepted for this event:

  • Marshall-Olkin Extended Zipf Distribution

    Authors: Marta Perez-Casany (Technical University of Catalonia), Aina Casellas (Technical University of Catalonia)
    Primary area of focus / application: Mining
    Keywords: Zipf distribution, scale-free network, skew distribution, discrete data analysis, data management, node degree.
    Submitted at 24-Apr-2013 10:30 by Marta Perez-Casany
    16-Sep-2013 11:20 Marshall-Olkin Extended Zipf Distribution
    Researchers from different areas such as ecology, engineering, social or computer networks, industry or physics use the Zipf distribution to model rank data, as well as data presented as frequencies of frequencies. This is because, many real data sets corresponding to counts exhibit a large probability at one, have a heavy-tail and show a linear behaviour in the log-log scale, which are the main characteristics of the Zipf distribution. For instance, this is the case of the following random variables: the number of links to web sites, the number of customers affected in electrical blackouts or the degree of nodes in a known network representation.

    Nevertheless, some data sets only show the linear behaviour for values larger than a given bound. When this happens, one needs to use a more flexible probability distribution. In this work the Zipf distribution is generalized by means of the Marshall-Olkin transformation. The resulting distribution, denoted as Marshall-Olkin Extended Zipf distribution (MOEZipf), is a two-parameter distribution that exhibits more flexibility in modelling the probability of the first positive integer values, while keeping the linear behaviour in the tail. Here the main properties of the MOEZipf family of distributions are presented, and several real data sets are analyzed. The results obtained prove the usefulness of the generalized model.
  • Dimensional Analysis and its Applications in Statistics

    Authors: Dennis Lin (Penn State University)
    Primary area of focus / application: Other: USA Session
    Keywords: Dimension Reduction, Data Analysis, Design of Experiment, Control Chart, Computer Experiment
    Submitted at 25-Apr-2013 22:01 by Dennis Lin
    17-Sep-2013 09:00 Dimensional Analysis and Its Applications in Statistics
    Dimensional Analysis (DA) is a fundamental method in the engineering
    and physical sciences for analytically reducing the number of
    experimental variables prior to the experimentation.
    The principle use of dimensional analysis is to reduce from a study
    of the dimensions of the variables on the form of
    any possible relationship between those variables.
    The method is of great generality.
    In this talk, an overview/introduction of DA will be first given.
    A basic guideline for applying DA will be proposed,
    using examples for illustration.
    Some initial ideas on using DA for Data Analysis and Design of
    Experiment will be discussed.
  • Selecting a D-optimal Follow-up Experiment among Candidate Choices

    Authors: David Edwards (Virginia Commonwealth University)
    Primary area of focus / application: Other: US invited session
    Keywords: Alias matrix, Bayesian inference, D-optimality, Follow-up design, Multiple criteria, Posterior predictive distribution
    Submitted at 26-Apr-2013 20:13 by David Edwards
    17-Sep-2013 09:20 Selecting a D-Optimal Follow-Up Experiment Among Candidate Choices
    Augmenting an initial screening experiment with additional runs is often needed to resolve ambiguities involving aliasing of effects. D-optimal augmentation is an efficient means of selecting a follow-up experiment that provides for precise estimation of a user-specified model by selecting runs such that the combined design maximizes the determinant of the information matrix. Furthermore, optimal design augmentation can offer greater flexibility with respect to run size and model specification than traditional followup strategies such as foldover and semifoldover. In this talk, we consider the use of multiple criteria for selecting among candidate D-optimal follow-up designs. For situations in which optimization is an eventual objective of experimentation, we also suggest a Bayesian approach to help select among candidate follow-up experiments through the computation of posterior probabilities that follow-up treatment combinations achieve some desirable quality level or meet required specifications.
  • Accelerated Destructive Degradation Test: Data Analysis and Test Planning

    Authors: Yili Hong (Virginia Tech), Caleb King (Virginia Tech)
    Primary area of focus / application: Other: US invited session
    Keywords: Lifetime prediction, Polymer degradation, Reliability, Robust design
    Submitted at 28-Apr-2013 14:58 by Yili Hong
    17-Sep-2013 09:40 Accelerated Destructive Degradation Test: Data Analysis and Test Planning
    Accelerated degradation tests are often used to obtain material reliability information in a timely manner. In tests where the measurement of the material property is destructive, only one measurement can be obtained for each test sample. Such test is referred to as accelerated destructive degradation test (ADDT). For polymeric materials, ADDT is often used to obtain long term property performance with temperature as the acceleration variable. The current industrial standard suggests a two-step approach for data analysis, which uses polynomials to interpolate the time to failure and uses the least square approach to fit the time-temperature relationship. An alternative approach is the ADDT data analysis approach which models degradation path as a function of time and temperature in one step. In this talk, we will discuss the pros and cons of both approaches. The ADDT model can provide the quantification of the uncertainty associated with reliability estimation. Based on the ADDT model, one can use simulations to study the performance of the standard analysis method. One can also develop statistically efficient and robust test planning procedure. We will discuss the possibility of conducting test with fewer samples under shorter time period but still with the same precision in the estimation.
  • Design and Sensitivity Analysis for Functional Inputs in Computer Experiments

    Authors: Jana Fruth (TU Dortmund University), Oliver Roustant (École Nationale Supérieure des Mines - Saint-Étienne), Sonja Kuhnt (TU Dortmund University)
    Primary area of focus / application: Design and analysis of experiments
    Keywords: computer experiments, functional input, sensitivity analysis, design of experiments
    Submitted at 29-Apr-2013 11:43 by Jana Fruth
    16-Sep-2013 12:35 Design and Sensitivity Analysis for Functional Inputs in Computer Experiments
    In time-consuming computer experiments, sensitivity analysis methods are largely applied to quantify the influence of input variables on the experiment. They are further used in variable reduction, model building and model validation among others. Especially popular are Sobol indices, which quantify numerically the influence of each input on the output. So far, these methods are reduced to scalar inputs, but more and more computer experiments allow for functional input parameters, i.e. settings in engineering processes that can be varied during the time of the experiment or parameters that vary in space. This talk presents ideas for a sensitivity analysis of functional inputs including design and graphical representation of the functional influences.

    Our specific aim is to not only get the sensitivity of the function as a whole, but to obtain a function of sensitivities that provides information on the influence of the different points (e.g. in time or space). We handle the huge dimension of this problem by exploring whole intervals of the functional domain at once. The input variables are, therefore, designed as piecewise constant functions, similar to B-splines of order 1. Then, we sequentially refine the interval size for interesting spaces of the functional domain. This results in a piecewise constant sensitivity for each functional input pointing out important and unimportant domain parts. Finally, the talk presents a current application in sheet metal forming with friction and blank holder force as functional inputs.
  • Identifying Locations and Scheduling of Hospitals and Pharmacies on Duty Aided Simulation in Isparta Downtown

    Authors: Halil Ibrahim Koruca (Suleyman Demirel University), Tuğba Özmen (Suleyman Demirel University), Merve Sari (Suleyman Demirel University)
    Primary area of focus / application: Modelling
    Keywords: Simulation, Health, Patient, Scheduling
    Submitted at 30-Apr-2013 19:12 by halil ibrahim koruca
    In health sector, treatment processes having short durations responding to the treatments shorten the healing time and improves patient satisfaction. Having limited hospital, doctor and pharmacy for a patient that gets ill at night, extends the time to reach a doctor and get medicine and causes queues in hospitals and pharmacies. The number of doctors and health staff, the travel time from hospital to pharmacy on duty, the number of pharmacies on duty and their location and the staff in pharmacies affect the process of the patient from the beginning to the end.
    The purpose of this study is decreasing the waiting time in hospitals and pharmacies, using resources more efficient and reducing the circulation time for patients. In this study, the number of people who visits the hospitals and pharmacies on duty in a month and the number of hospitals and pharmacies on duty are counted. According to these data current system simulation was modeled in ARENA. Alternative scenarios are developed for working schedules and locations of hospitals and pharmacies on duty in Isparta Downtown and it is compared with current state. The answers of the farther questions is tried to determine from the results that how many doctors and health staff is needed in which hospital, the locations of the pharmacies and which days which pharmacies should stay open.