ENBIS: European Network for Business and Industrial Statistics

My abstracts

The following abstracts have been submitted for this event:

• Fuzzy Conceptual Clustering
  
  Authors: Petra Perner
  Affiliation: Institute of Computer Vision and Applied Computer Sciences, IBaI Submitted at 3-Mar-09 09:55 by Petra Perner
  Accepted (view paper) Classical clustering methods only create clusters but do not explain why a cluster has been established. Conceptual clustering methods built cluster and explain why a set of objects confirm a cluster. Thus, conceptual clustering is a type of learning by observations and it is a way of summarizing data in an understandable manner. In contrast to conventional hierarchical clustering methods, conceptual clustering methods built the classification hierarchy not only based on merging two groups. The algorithmic properties are flexible enough in order to dynamically fit the hierarchy to the data. This allows incremental incorporation of new instances into the existing hierarchy and updating this hierarchy according to the new instance.
  We present a conceptual clustering methods based on a fuzzy similarity measure.To our knowledge there exists no algorithm for Fuzzy conceptual clustering yet.
  We apply this kind of clustering method for novelty detection in an on-line process for high-content analysis of cell-based assays.
  We apply this kind of clustering method for novelty detection in an on-line process for high-content analysis of cell-based assays.
• An incremental approach to efficient experimentation
  
  Authors: Matthew Dodson, Lois Dodson and Rene Klerx
  Affiliation: Design of Experiments Submitted at 5-Mar-09 09:58 by Rene Klerx
  Accepted (view paper) Experimental trials are often expensive and time consuming. When prototypes or durability tests are required, an experimental trial can easily cost several thousand euros. This paper will present an approach to minimising the number of trials when conducting an experiment. A full factorial experiment provides an estimate of effect for all main effects and interactions. A more efficient approach is to start with a Resolution III design, and separate confounding between interactions and main effects with additional trials. This methodology will be demonstrated with a five factor experiment.
• Who stole SPC?
  
  Authors: Jonathan Smyth-Renshaw
  Submitted at 5-Mar-09 22:10 by Jonathan Smyth-Renshaw
  Accepted (view paper) I wish to give a talk even a short half day course on the correct use of SPC. It is clear from my work as a consultant that many businesses do not understand the correct way to use SPC to both understand and drive business improvement. The technique first used in 1924 has been lost in a maze of computer code. A return to basics is long over due and it is this I wish to aid in my presentation/course.
• Integrating Operational and Financial Risk Assessments
  
  Authors: Silvia Figini Ron Kenett Juan Tomas Sayago Paolo Giudici
  Submitted at 19-Mar-09 15:10 by SILVIA FIGINI
  Accepted (view paper) Operational risks are typically classified into hardware, software, interface, network and security related events. Assessing operational risks involves merging data from different sources such as system logs, call centre records, technical service data bases and customer complaints. Financial risks are reflected by balance sheets, performance trends and expert opinions. In this paper we will present a new approach to integrate the two domains and provide a combined assessment. Such an assessment considers specific companies and their segment. We will present our approach in the context of a company serving customers with installed Private Branch Exchanges providing telecommunication capabilities, including Voice over IP. Each customer is assessed in terms of it's operational risks, in a specific way and in terms of the segment it belongs. A similar assessment is conducted with respect to its financial risks. The models we propose include a customer specific and a segment specific contribution. Such models allow decision makers determine the proper business models for serving their customers.
• EWMA Control Charts for Monitoring Binary Processes with Applications to Medical Diagnosis Data
  
  Authors: Christian H. Weiß, University of Würzburg, Institute of Mathematics, Department of Statistics, Germany. Martin Atzmüller, University of Würzburg, Institute of Computer Science, Department of Artificial Intelligence, Germany.
  Submitted at 23-Mar-09 16:15 by Christian Weiß
  Accepted (view paper) The statistical control of attribute data processes is an area of emerging interest both in research and professional practice. Often one is concerned with binary attribute processes. The traditional example for such binary quality characteristics is the case, where the process describes the result of an inspection of an item, taking the value 1 iff the item was defective or nonconforming. The presented research, however, was motivated by a non-manufacturing example: medical diagnosis data.
  
  To monitor a binary process, exponentially weighted moving average (EWMA) control charts in different variations are proposed. The ARL performance of the EWMA approach both with standard and with skewness-corrected 3-Sigma control limits is investigated and design recommendations are derived. The proposed EWMA control charts are applied to medical diagnosis data taken from the diagnostic expert system SonoConsult in order to provide a semi-automatic component for monitoring the documentation behavior of different examiners.
• Multivariate Control Charts
  
  Authors: Marianne Frisén
  Affiliation: Statistical Research Unit, Department of Economics, University of Gothenburg, Sweden Submitted at 5-Apr-09 09:02 by Marianne Frisén
  Accepted (view paper) Multivariate control charts are of interest in industrial production as it enables the monitoring of several components. Monitoring of the total quality of an assembled product is one example. Recently there has been an increased interest also in other areas such as regional public health control and financial transaction strategies. Reviews are found in Sonesson and Frisén (2005) and Frisén (2009).
  Multivariate counterparts to the univariate Shewhart, EWMA and CUSUM methods have earlier been proposed. A review is given with respect to how suggested methods relate to general statistical inference principles.
  Optimality is usually hard to achieve, and even to define, in multivariate problems. This is so also for multivariate surveillance. Multivariate surveillance problems can be complex. How-ever, the sufficiency principle allows important reductions of some important classes of mul-tivariate surveillance problems as seen in Frisén, Andersson and Schiöler (2009b). The spe-cial challenges of evaluating multivariate surveillance methods (see Frisén, Andersson and Schiöler (2009a)) will be described.
  References
  Frisén, M. (2009). Principles for Multivariate Surveillance, to appear In Frontiers in Statistical Quality Control, H.-J. Lenz, and P.-T. Wilrich (eds).
  Frisén, M., Andersson, E.and Schiöler, L. (2009a). Evaluation of Multivariate Surveillance. In Research Report 2009:1: Statistical Research Unit, Department of Economics, University of Gothenburg, Sweden.
  Frisén, M., Andersson, E.and Schiöler, L. (2009b). Sufficient reduction in multivariate surveillance. In Research Report 2009:2: Statistical Research Unit, Department of Economics, University of Gothenburg, Sweden.
  Sonesson, C.and Frisén, M. (2005). Multivariate surveillance. In Spatial surveillance for public health, A. Lawson, and K. Kleinman (eds), 169-186. New York: Wiley.