ENBIS-8 in Athens
21 – 25 September 2008
Abstract submission: 14 March – 11 August 2008
The following abstracts have been accepted for this event:
While normal and half-normal plots together with Lenth’s method are used extensively in the analysis of regular two-level design and their fractions, these methods may be of little help in the analysis of non-regular two-level designs like Plackett-Burman designs for instance. The reason why is that the aliasing between main effects and higher order effects is rather complex. However, when only a few higher order interactions are active, many non-regular designs have a very structured alias pattern. In this talk we demonstrate how scatter-plots, cluster analysis and a method based on orthogonalized scatter-plots may give valuable information about active effects and also be supportive for other analysis methods for these designs.
Statistical process control is widely used to monitor the quality of a process or of the final product of it. Control charts help the practitioners to identify assignable causes so that the state of statistical control can be achieved. Intuitively, in the event of having an undesirable shift in the process, a control chart should detect it as quickly as possible and give an out-of-control signal. In this article, we introduce a new Wilcoxon-type rank sum control chart, which due to its nonparametric nature does not require the assumption of any specific probability distribution for the underlying process(measurements). More specifically, while the process is in-control, a reference sample is taken from it and the LCL and UCL of the control chart is determined by exploiting appropriate order statistics of the reference sample. Then, test samples are drawn independently of each other (and also of the reference sample) and the decision whether the process remains in-control or not, is based on the sum of the ranks of the test sample observations which lie between the control limits. Exact formulae for the alarm rate, the run length distribution and the average run length (ARL) are all derived. Tables are provided for the implementation of the chart for some typical false alarm rates and numerical comparisons are presented between it and other non-parametric Shewhart-type charts.
The traditional approach to maintain professional copiers is to send a service technician when there is a permanent error, along with planned regular visits. For a case study within the IRIS project (Intelligent Remote Infrastructure Ser-vice project), a regional industrial project with several partners, we investi-gated the possibilities to use information sent by copiers through the Internet to implement maintenance strategies based on condition monitoring. The in-formation consisted of daily dumps sent to an FTP server containing informa-tion on several internal counters, as well as generated error and warning mes-sages.
We discuss the practical problems that arose in data conversion from hierar-chical XML to sensible statistical formats, data inconsistency and selection of meaningful variables. Moreover, we will share our experiences in choosing appropriate software and statistical techniques.
Variability is an important aspect of SRAM memory cell design. Failure prob-abilities of Pfail≤10^-10 have to be estimated through statistical simulations using time consuming circuit simulators. Accurate statistical techniques such as Importance Sampling Monte Carlo simulations are essential to accurately esti-mate such extremely small failure probabilities. We show that a simple form of Importance Sampling is suffices for simulating Pfail≤10^-10 for important per-formance characteristics like the Static Noise Margin, Write Margin and Read Current. For the SNM, a new simple technique is proposed that allows extrapolating the SNM distribution based on a limited number of trials.
To remain competitive in the market place of consumer products, industrial companies start to realize that it is essential to develop products that match the requirements of the modern and critical user. Therefore, user tests play an important role in developing consumer products in industry. Often, these tests are set up and analyzed using techniques from Design of Experiments. Typically, the blocking principle is used and the users are considered as blocks. A common problem is that it is not possible, for each test person, to undergo all treatments; for instance, because the person would get tired or bored. Therefore, the test design is chosen to be a balanced incomplete block design. A standard assumption in this set up, and in the analysis of the experimental data, is that the interaction between blocks and treatments is absent. However, it is our experience from various user tests that these interactions cannot be ignored beforehand. We therefore studied the consequence of block-treatment interactions for the analysis, by ANOVA and linear mixed models, of data from balanced incomplete block designs in user tests. The results from this study will be presented and conclusions will be given.
Websites form a major gateway into services offered by niche consultancies. Websites are the first stop for customers to find information about the services offered. This experience can be vital in the customer forming his first impressions about the organisation. With web channel being the most slippery channel, a potential customer can easily be lost to competitors. Website design is therefore extremely important. Web usability and design factors which emotionally appeal to the customers help the formation of “sticky” relationships with customers. Facilities which provide metrics for web analytics are freely available and can be helpful if used appropriately. Feedback from customers is a fundamental source of information to guide design and assess effectiveness of a website. Customer-centred design is popular in most conventional businesses such as automobiles and food packaging. In the present paper, we propose a customer-centred design strategy for website design to meet customers’ emotional needs as users and patrons. Kansei Engineering is a methodology which captures emotional response to products and services. An administrative change provided an opportunity to review the ISRU website and to redevelop it. ISRU is a self-financing unit within the School of Mathematics and Statistics, Newcastle University. The statistical services it offers are of general applicability but nevertheless seek to satisfy a niche market. A baseline summary was obtained to compare with post implementation results. The well established I*PROMS website provided access to examples of historic and current web analytic data which were used to develop metrics for assessing effectiveness in the particular case of a niche technical consultancy unit. The paper reports the project which was carried out using Six Sigma project methodology. Websites cannot remain static but must be responsive to disruptive innovation and other vectors of change. Website design is put in a historical context with the aim of tracing future requirements for change.