ENBIS-13 in Ankara
15 – 19 September 2013
Abstract submission: 5 February – 5 June 2013
The following abstracts have been accepted for this event:
The goal of this project* was to predict the reliability of metal parts of a consumer product**, based on fatigue testing. It was assumed that fatigue is one of the failure mechanisms of the product. In the test, the metal parts were stressed by pressing at a specific position with a specific force. Different levels of force were used on different parts. The pressing was repeated until the part failed, or censored after 10^6 times. The translation of these test results to reality was challenging. For describing the real use of the product, empirical distributions of number of pushes, force and position of the force were available. These real life distributions were combined to model the lifetime. A force at any position in real use was transformed to a different force on the test position, keeping the stress level in the metal constant. This translation was obtained by computer simulation. Also, the reliability prediction needed to be consistent with theory of metal fatigue, for example, following the Wöhler curve. Using a Weibull distribution and by Monte Carlo simulation, the lifetime and failure rates were predicted. As a final step, the reliability of a single part was generalized to the reliability of a total system, with a total of 27 parts.
*This was a consultancy project of the company CQM.
**Due to the confidential nature, the details of the product are not given.
The increasing complexity of multi-criteria group decision making problems requires the use of more ﬂexible approaches to prioritize the alternatives. Methodological developments have been aroused in the literature in order to overcome the limitations of classical aggregation methods in group judgements. Bayesian methods can deal with missing or incomplete data using data augmentation techniques. The objective of this study is to show how Bayesian approaches provide more effective and practical group decision mechanisms to both practitioners and researchers. The methodology is illustrated using the Analytic Network Process in a case study.
16-Sep-2013 11:40 Application of Bayesian Approaches in Multi-criteria Group Decision Analysis
Application of Bayesian Approaches in Multi-criteria Group Decision Analysis
Zeynep Filiz Eren-Dogu (Mugla Sitki Kocman University), Can Cengiz Celikoglu (Dokuz Eylul University)
Primary area of focus / application:
Keywords: Multi-criteria group decision making, Bayesian approach, Analytic Network Process, Negotiations
Submitted at 30-May-2013 14:53 by Zeynep Filiz Eren Dogu
In recent years, improvements in technology and increases in the number of investors, allowed production of the Turkish movies with more diverse topics and more budgets. This leads to the movie industry to expand and become an income and expenditure resource for larger populations. More and more movies are being produced every year resulting in an increase in competition. Although, in Turkey, TV series have big share in the market by exports and advertisement profits, movies still create significant transactions. A successful movie can be considered as the one which makes higher revenues than its costs, when the gap is higher movie is considered more successful. The success of a movie, in terms of the revenue and number of viewers, depends on many factors such as the quality of production, marketing, cast etc. In this study, the effect of release date and genre on the total revenue and number of viewers is investigated for Turkish movies. The corresponding data is analyzed between 2005-2013 to see if there exists any pattern or relationship among these variables. The preferred times of the year for the release date and the effect of the holidays on the opening week revenues are also investigated.
16-Sep-2013 16:05 A Research on Turkish Movie Industry: Preferred Released Dates
A Research on Turkish Movie Industry: Preferred Released Dates
Guldal Guleryuz (Hacettepe University), Ebru Yuksel (Hacettepe University)
Primary area of focus / application:
Keywords: movies, release date, correlation analysis, descriptive statistics
Submitted at 30-May-2013 15:04 by Guldal Guleryuz
Soccer is a big part of both entertainment and financial industries in Turkey. It is also the sport which receives the most attention from the society. The impact of derby games between major soccer games on stock exchange and production indexes have been analyzed in the literature extensively. Those games affect the expenditure, stimulate consumption and create economic and financial activity as well as increasing the traffic and safety/security measures. In this study, we analyze the impact of major soccer teams’ matches on various financial performance indicators in Turkey. We investigate the importance of major soccer events which affect a lot of people emotionally on asset pricing in accordance with behavioral finance.
Multivariate statistical process monitoring (SPM) methods often rely on some form of Principal Component Analysis (PCA) in order to simultaneously deal with the high-dimensionality and time-dependency features of real world processes. These methodologies can be divided into two types of approaches: non-adaptive and adaptive. Non-adaptive approaches include the classic PCA approach and Dynamic PCA for data with auto-correlation, while Recursive PCA and Moving Window PCA, derived for non-stationary data, are adaptive. To our knowledge, there is no comprehensive investigation indicating when each method is more appropriate or examining their relative performance. Therefore, a comparison of the performance of these methods on different process scenarios is performed, and guidelines are outlined on the selection of the most appropriate monitoring strategy for each scenario. The process scenarios are drawn from a selection of simulated and real data cases with high-dimensionality and time dependence properties. These characteristics make many of the commonly used methods, which assume stationarity and nonautocorrelated variables, unsuitable for use. This stresses the need to properly select the correct monitoring methodology. The selection of parameter values for PCA methods, such as the number of lags for DPCA or the forgetting factor for RPCA, is another crucial factor for their successful implementation, and guidelines for doing so are also discussed. Additionally, approaches for simulating data with desirable properties for testing the methods are covered. Finally, we consider the challenges faced in the modeling of time-dependent data, and highlight areas of possible further research for each of the PCA investigated approaches.
The time it takes to perform a particular service is variable. Examples include the time it takes a medical professional to see a patient, the time it takes to perform equipment maintenance, and the time duration of a customer service call. The variance at any point in time is the sum of the variances of the previous actions. There is a cost for waiting for both the server and the service provider. A mathematical model is presented to determine schedules that optimise the total system cost.
17-Sep-2013 09:20 Financial Optimisation of Appointment Scheduling
Financial Optimisation of Appointment Scheduling
Rene Klerx (SKF Group Six Sigma), Jonathan Slater (Lawrence Technological University), Paolo Re (SKF Group Six Sigma), Bryan Dodson (SKF Group Six Sigma)
Primary area of focus / application:
Keywords: Reliability, Statistical Optimisation, Maintenance, Six Sigma
Submitted at 30-May-2013 16:36 by Rene Klerx