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:

  • Using Data Mining Modelling for Process Improvement on Polyester Foams Type – a Case Study

    Authors: Silvio Costa (Flex 2000) Nuno António (StatSoft)
    Affiliation: Flex2000/StatSoft
    Primary area of focus / application: Modelling
    Keywords: Data mining , foam , process improvement , model selection , model deployment
    Submitted at 20-Jun-2011 12:36 by Nuno Antonio
    Accepted
    6-Sep-2011 16:05 Using Data Mining Modelling for Process Improvement on Polyester Foams Type – a Case Study
    The slabstock flexible foam industry Manufacturing Processes can be influenced by a large number of factors related to raw materials quality, physical attributes, formulations, engineering decisions. Data Mining algorithms can play an important role on process development because they show more flexibility to fit even very non-linear data while keeping a satisfactory predictive capacity.
    This project included an holistic approach to the process improvement of the polyester flexible foam which included several steps:
    - Merging data, with many different formats, stored on different sources (certificates of analysis, environmental conditions, machine settings, chemical formulations);
    - Performing variable screening as well as checking for variable redundancies in order to prevent collinearity problems;
    - Fitting data mining models to predict the probability of obtaining a non-conformity using several data mining algorithms;
    - Evaluating the models quality through V-fold cross-validation and residual analysis;
    - Using a multi-criteria approach to select the best model in order to balance predictive power with the usage of actionable variables;
    - Deploying the final model on the real production environment as a management tool for the Production department of Flex2000.
    This project is a joint work of Flex2000 Quality department and StatSoft Ibérica analytical team.
  • Statistical Prediction of Road Traffic. From Data to Models and viceversa

    Authors: G. Allain, F. Gamboa, Ph Goudal, JN Kien, JM Loubes
    Affiliation: Institut de Mathématiques de Toulouse
    Primary area of focus / application: Modelling
    Keywords: Road , Traffic , Statistics , Modelling
    Submitted at 20-Jun-2011 15:52 by Gamboa Fabrice
    Accepted
    5-Sep-2011 12:35 Statistical Prediction of Road Traffic. From Data to Models and viceversa
    Session: Business and Industrial Statistics in France


    One of the main needs of any driver is to get an accurate knowledge on his travel time. It requires a blend of good intuition and mathematical analysis.

    Prediction of traveling time have become a real challenge. Huge traffic jams strike every day road networks leading to our cities and suburbs, yielding terrible costs in time and money.

    Of course, traffic models based on physics have been developed. However, they deal with too many parameters and so cannot be used to predict in real life travel time. As a matter of fact, these models provide only predictions restricted to very few points of the network.

    Our work aims at giving prediction time algorithm for any car travel. It has begun in 2001. It received an ANVAR award in 2004 and a grant fund from ANR in 2006. It goes on through a cooperation with the private company MEDIAMOBILE.

    Modelling the drivers’ behaviors

    Our approach relies on the hypothesis that road traffic evolution can be summarized by two components. One is linked to periodical behaviors of users on the network. This component is recurrent on a daily basis and can sometimes be predicted deterministically. The other component is the result of dynamic interactions in the physical flow of vehicles on the road. This component may describe the propagation of a specific traffic congestion along the network, or more generally, the various reasons why traffic evolution on a specific day may locally diverge from the deterministically predicted behavior.

    Prevision models in each point of the network are composed of an alphabet of typical daily speed evolution profiles built upon a four year historical database on more than 10000 points. This alphabet of typical profiles is then used to recognize the component in the mixture for the current day by using observed speeds and calendar characteristics of the day.

    The model mentioned above brought us to solve several inter-independent problems : finding a dissimilarity between speed curves embedding traffic characteristics, coming up with clustering methods and last but not least defining a model identification method yielding a good trade-off between generality and accuracy. We used complex statistical methods as non-linear regression methods and support vector machines, while proposing automatized, modular and efficient methods which also meet the industrial constraints we face.

    This alphabet of typical profiles is then used to recognize the most likely evolution for the current day by using observed speeds and calendar characteristics of the day. Our algorithm estimates short and medium term average vehicle speeds on all the road network, thus enabling to provide the driver with an accurate travel time prediction of his trip. This system is used on MEDIAMOBILE’s website <http://www.v-trafic.com/#vtactic>.

    Extending the coverage

    In a second time, we used data gathered sporadically from the analysis of in-traffic vehicles positions. These data yield industrial interest because they can be collected globally on the network contrary to traditional data collected from counting station made of electromagnetic loops built in the road. Accurate traffic state is reconstructed through pioneering statistical methods. Those methods are based on time-space analysis of graphs and model both vehicle physics and the road network geometry.
  • A non-parametric approach to measuring the shape property of nonlinear profiles

    Authors: Stelios Psarakis*, Javier Cano**, Javier M. Moguerza**, Athanasios N. Yannacopoulos*
    Affiliation: *Department of Statistics, Athens University of Economics & Business **Department of Statistics & Operations Research, Rey Juan Carlos University
    Primary area of focus / application: Process
    Keywords: Profile monitoring , shape property , statistical process control , nonparametric procedures
    Submitted at 21-Jun-2011 11:26 by Stelios Psarakis
    Accepted (view paper)
    5-Sep-2011 17:15 A non-parametric approach to measuring the shape property of nonlinear profiles
    The quality of a process or product can be characterized by a functional relationship between a response variable and some explanatory variables. In this work we develop nonparametric procedures for the study of non-linear profiles, that is, nonlinear noisy functional relationships between variables. In particular, we focus on the shape property of profiles. To this aim, we design a metric based on the solution of an optimization problem. In addition, we show that the problem is well-posed from a theoretical point of view. Finally, we illustrate the performance of the proposal with numerical examples.
  • Six Sigma in Healthcare

    Authors: Blanton Godfrey
    Affiliation: North Carolina State University
    Primary area of focus / application: Six Sigma
    Keywords: health care , Six Sigma , Lean Six Sigma , quality
    Submitted at 21-Jun-2011 17:35 by Blanton Godfrey
    Accepted (view paper)
    6-Sep-2011 09:00 Six Sigma in Healthcare
    In the 1980s Motorola pioneered the methods now collectively known as Six Sigma Quality. In the following years companies such as General Electric and Honeywell expanded these concepts and received much favorable press for the stunning results they achieved. Six Sigma quickly spread to thousands of manufacturing companies throughout the world. It was not until the early 2000s that we began to see implementation of these ideas and methods in leading healthcare organizations. Now many leading healthcare organizations are using many of the methods of Lean and Six Sigma both to improve quality of care and to drive down costs. We will summarize the basic concepts of Lean Six Sigma and focus on their application in leading healthcare organizations.
  • Six Sigma in Healthcare

    Authors: Blanton Godfrey
    Affiliation: North Carolina State University
    Primary area of focus / application: Six Sigma
    Keywords: health care , Six Sigma , Lean Six Sigma , quality
    Submitted at 21-Jun-2011 17:35 by Blanton Godfrey
    Accepted (view paper)
    6-Sep-2011 09:00 Six Sigma in Healthcare
    In the 1980s Motorola pioneered the methods now collectively known as Six Sigma Quality. In the following years companies such as General Electric and Honeywell expanded these concepts and received much favorable press for the stunning results they achieved. Six Sigma quickly spread to thousands of manufacturing companies throughout the world. It was not until the early 2000s that we began to see implementation of these ideas and methods in leading healthcare organizations. Now many leading healthcare organizations are using many of the methods of Lean and Six Sigma both to improve quality of care and to drive down costs. We will summarize the basic concepts of Lean Six Sigma and focus on their application in leading healthcare organizations.
  • Forecasting: a practitioner’s view

    Authors: Luis M. Artiles Martinez
    Affiliation: Accenture Plc
    Primary area of focus / application: Consulting
    Keywords: forecasting , analytical consultancy , supply chain , dynamical systems , highly dimensional data
    Submitted at 21-Jun-2011 17:37 by Luis M. Artiles Martinez
    Accepted
    6-Sep-2011 15:25 Forecasting: a practitioner’s view
    Several issues arrive when handling real data: from one extreme with missing data to the other extreme with massive and highly dimensional data. In terms of data structure complexity comes in several forms when modeling dynamical systems: noise, censoring, intermittence, dynamic variance, multidimensional and highly correlated signals, complex data structures. We shall discuss those issues while making an overview of current practices in Analytical Consultancy today, specifically in Supply Chain.