ENBIS-14 in Linz

21 – 25 September 2014; Johannes Kepler University, Linz, Austria Abstract submission: 23 January – 22 June 2014

My abstracts

 

The following abstracts have been accepted for this event:

  • Risk Valuation for European Electricity Trading with Hourly Pricing Forward Curves

    Authors: Martin Rainer (ENAMEC Institut und FZRM Univ. Würzburg, SYNECO Trading GmbH, München)
    Primary area of focus / application: Modelling
    Keywords: Pricing forward curves, Electricity markets, Risk valuation, Value-at-risk, Risk management
    Submitted at 2-Jul-2014 17:47 by Martin Rainer
    Accepted
    23-Sep-2014 09:20 Risk Valuation for European Electricity Trading with Hourly Pricing Forward Curves
    The hourly pricing forward curve (PFC) is a key input for pricing and risk management of European electricity markets trading.

    In these markets the spot prices are quoted for each hour of a day. Furthermore, traded forward instruments for base and peak load provide a basis of defined forward risk factors. It is common practice to derive hourly PFC (h-PFC) curves from the latter and estimated daily and hourly profiles. The details of this construction may genuinely
    affect the correlation structure of the h-PFC.

    An additional problem is the existence negative market prices, because it contradicts the standard assumptions of conventional value-at-risk (VaR) methodology. This in particular requires a modification of the VaR-concept of risk management.

    Several approaches for this purpose are discussed from both perspectives of theoretical foundations and practical viability.
  • Structured Data Analysis with Regularized Generalized Canonical Correlation Analysis

    Authors: Arthur Tenenhaus (Supelec)
    Primary area of focus / application: Mining
    Keywords: Regularized generalized canonical correlation analysis, Structured data analysis
    Submitted at 1-Aug-2014 16:49 by Arthur Tenenhaus
    Accepted
    23-Sep-2014 17:30 Best Manager Award (Tim Davis) and Young Statistician Award (Arthur Tenenhaus: Structured Data Analysis with Regularized Generalized Canonical Correlation Analysis)
    In contrast to standard data that is structured by a single individuals*variables data matrix, structured data are characterized by multiple and heterogeneous sources of information, interconnected, potentially of high dimensions. In addition, each source of information may also have a complex structure (e.g. tensor structure). The need to analyze the data by taking into account their natural structure appears to be essential but requires the development of new statistical techniques that constitute the core of my current research for many years. More specifically, I will present a general framework for structured data analysis through Regularized Generalized Canonical Correlation Analysis.
  • Multivariate Latent Variable Models: Misconceptions and Industrial Applications

    Authors: John F. MacGregor (McMaster University and ProSensus, Inc.)
    Primary area of focus / application: Modelling
    Keywords: Multivariate latent variable models, Industrial application, Projection to latent structure
    Submitted at 2-Aug-2014 14:38 by John MacGregor
    Accepted (view paper)
    23-Sep-2014 11:45 George Box Award: John F. MacGregor. Award talk on "Multivariate Latent Variable Models: Misconceptions and Industrial Applications"
    Multivariate latent variable regression models based on Projection to Latent Structures (PLS) offer many unique properties that make them well suited for the analysis of historical industrial data. These properties, which include uniqueness, interpretabilty, handling missing data, causality in the latent variable space, all arise from the fact that they simultaneously provide latent variable models for both the regressor space (X) and the response space (Y).
    These properties allow for many interesting industrial applications. Several of these will be illustrated in this presentation for industrial batch processes, including: (i) the analysis and interpretation of batch production data, (ii) the optimization of batch operations, and (iii) the monitoring and control of batch processes.
  • A Bootstrap-Based Correction Method for the Hotelling’s Multivariate Control Chart for Serially Dependent Data

    Authors: András Zempléni (Eötvös Loránd University, Budapest), László Varga (Eötvös Loránd University, Budapest)
    Primary area of focus / application: Process
    Keywords: Block bootstrap, Effective sample size, Hotelling’s control chart, Serial dependence
    Submitted at 6-Aug-2014 14:00 by András Zempléni
    Accepted
    24-Sep-2014 10:35 A Bootstrap-Based Correction Method for the Hotelling’s Multivariate Control Chart for Serially Dependent Data
    In a recent paper [1] we introduced a model-based method for considering the dependence between wind-speed observations in a bivariate setup. Now we show its potential for financial data (daily log-returns for foreign exchange rates), where we try to signal changes in the behaviour of the process by the well-known Hotelling chart. The data are not independent (let us denote the sample size by n) and the usual correlation-based methodology is not applicable, because the dependence in not linear.

    The effective sample size n_e is defined as the size of the independent sample, for which the trace of the covariance matrix is the same as the one of the empirical data. The critical values are then the usual critical values of the Hotelling’s chart, but based on n_e observations instead of the original n.

    The block bootstrap methodology (see [2]) is an effective tool for resampling serially dependent data. However, the determination of the block size is by far not obvious. We propose a method, where n_e is determined by two approaches: one is model-based (vector-autoregressive or GARCH models are fitted to the data), the other is by the block bootstrap, where the block size is chosen, for which the simulated trace of the covariance matrix is the nearest to the one given by the model. This block size may then be used for simulations, e.g. for bootstrap confidence regions.

    References
    [1] P. Rakonczai, L.Varga and A.Zempléni (2014): Copula fitting to autocorrelated data, with applications to wind speed modeling. Annales Univ. Sci. Budapest, Sect. Comp., to appear.
    [2] S. N. Lahiri (2003): Resampling Methods for Dependent Data, Springer.
  • DoE Empowerment

    Authors: Stefanie Feiler (AICOS Technologies Ltd.), Philippe Solot (AICOS Technologies Ltd.)
    Primary area of focus / application: Design and analysis of experiments
    Keywords: DoE , Software, STAVEX, DoE propagation, DoE education
    Submitted at 8-Aug-2014 16:25 by Stefanie Feiler
    Accepted (view paper)
    22-Sep-2014 16:00 DoE Empowerment
    The use of DoE (statistical design of experiments) is best practice in product and process development and optimisation.

    However, many researchers still show a certain reluctance to employ DoE.
    They judge the method to be complicated, suspect that they would be forced to perform too many experiments, or believe that it can only be mastered by statistics experts.

    The strength of the DoE expert system STAVEX is its user-friendliness.
    Users are guided in selecting an appropriate design for their specific situation, in the model analysis, and in deciding on the next experimental step. This allows the tool to be (correctly) used by "standard" researchers, without having to rely on the help of a statistician. We illustrate the various aspects with real-life examples.
  • The ReNewTown Project and the Importance of (Self-)Assessment

    Authors: Irena Ograjenšek (Faculty of Economics, University of Ljubljana)
    Primary area of focus / application: Modelling
    Keywords: ReNewTown, Lessons learned, Central Europe Programme, (Self-)assessment
    Submitted at 15-Sep-2014 11:34 by
    Accepted
    The project with the title New Post-Socialist City: Competitive and Attractive (in short the ReNewTown project: http://www.renewtown.eu), which has been implemented through the Central Europe Programme co-financed by the European Regional Development Fund, focuses on reduction of disparities in quality of post-socialist urban environment (not only by positive transformation of residential landscapes but also by improved quality and accessibility of public spaces, increased provision of cultural and social events, increased support of entrepreneurial initiatives, etc.). To this end several model approaches have been identified and four pilot actions implemented in post-socialist cities from four different Central and Eastern European countries (Czech Republic, Poland, Slovakia and Slovenia). This paper demonstrates the importance of (self-)assessment in all phases of pilot actions' implementation: its framework, motivation, obstacles, and lessons learned.