Logo des Repositoriums

Auflistung nach Autor:in "De Baets, Bernard"

1 - 3 von 3
  • Konferenzbeitrag
    A Lattice-Theoretic Approach to Computing Averaged Ranks Illustrated on Pollution Data in Baden-Württemberg
    (Environmental Informatics and Systems Research, 2007) De Loof, Karel; De Baets, Bernard; De Meyer, Hans; Brüggemann, Rainer
    In the paper an algorithm for calculating the ranking distribution and the averaged ranking probability (averaged rank) of each element of a poset that avoids the enumeration of all linear extensions and relies on the lattice of ideals representation of the poset is presented.
  • Konferenzbeitrag
    Directional Adjustment of Average Rankings of Two Posets: an Application to Herb Layer and Moss Layer Pollution in Baden-Württemberg
    (Environmental Informatics and Systems Research, 2007) Rademaker, Michael; De Baets, Bernard; De Meyer, Hans; Brüggemann, Rainer
    In the paper two procedures for Directional adjustment of average rankings of two posets: an application to herb layer and moss layer pollution in Baden-Württemberg are disscussed.
  • Konferenzbeitrag
    Partial Order Concepts in Ranking Environmental Chemicals
    (Environmental Informatics and Systems Research, 2007) Brüggemann, Rainer; Voigt, Kristina; Sørensen, Peter B.; De Baets, Bernard
    With the discussion around REACH the interest in ranking of chemicals is increased. In this paper we take a statistical point of view in evaluation and ranking of chemicals. The data matrix with relevant properties of the chemicals plays the central role. Its evaluation is performed by simple concepts of partial order (Hasse diagram technique (HDT)). Several extensions are discussed: First the concept of METEOR (METhod of Evaluation by ORder Theory), then the approach by fuzzy concepts, combined with partial order, and finally the DLPO-concept (Double Layer Partial Orders). DLPO is a novel method where attributes are themselves combined by partial order. The main idea is to avoid the bias due to a functional combination of attributes (for example by weighted sums as in METEOR). Whereas HDT does not need any information beyond the data matrix, all other methods discussed so far, need additional knowledge.