this paper: an algorithmic extension of the crisp relevance test, a Bootstrap fuzzy relevance test and an asymptotic fuzzy relevance test. The results of the Bootstrap fuzzy relevance test are very good, but the high computing time makes its application only practicable for a small number of data samples. The asymptotic fuzzy relevance test supplies good results for a higher number of data samples. The algorithmic extension of the crisp relevance test tends to calculate too large condence intervals, but has the smallest computing time. The employment of the three relevance tests will depend on the respective application. For high dimensional search spaces with a multitude of relevant rules, the algorithmic extension is acceptable, especially, if for each input and output variable several trapezium fuzzy sets are reasonable. In the other cases, the higher eoeort of the fuzzy relevance tests can remunerate. The calculation of estimators and condence intervals on fuzzy data is also meaningful for other test and rating strategies, e.g. the results can be directly used for the method 'Condent Normalized Hit Rate' of Jessen and Slawinski [9]. Acknowledgement

The paper presents a new approach to the automatic data-based generation of fuzzy rules. This is based on a tree-oriented rule induction algorithm and rule pruning. The hypothesis generation applies a set of measures for evaluation of fuzzy rules with respect to approximation quality, importance, clearness etc. In order to improve #exibility and interpretability linguistic hedges are used to create derived linguistic terms.

Most approaches for the data-based design... this paper is to construct an appropriate criterion for evaluating partitions or MBFs and to search for the optimal value. The algorithm bases on information-theoretical measures [8] and additional modifications to guarantee the interpretability and practical acceptance of the solutions. Because information-theoretical measures assume a discrete partition of input and output variables, new approaches to represent fuzzy membership values as a possibility distribution are introduced. The aims of this paper...

The paper presents new strategies for testing and rating the relevance of rules in the Fuzzy--ROSA (Rule Oriented Statistic Analysis) method for data based rule generation. Specific characteristics and differences between the proposed strategies are pointed out. Keywords: Fuzzy systems, rule based modeling, data based modeling, relevance test and rating, Fuzzy--ROSA method 1 Introduction Modeling of a given process can be carried out either theoretically or empirically. The theoretical approach is based on a theory and existing knowledge about the process. The empirical approach uses measured input/output data. The Fuzzy--ROSA (Rule Oriented Statistic Analysis) method is an empirical approach using fuzzy--if--then--rules to describe the observed behaviour of a process [1, 2]. The if--then--rules have the form IF p k THEN c k (1) with k indicating the k--th rule. The premise part p k of the rule is a statement on the input vector x and the conclusion part c k is a statement on ...

Abstract: Corporate bankruptcy prediction is a usual problem found in financing and management. In this paper a new approach is proposed based on Machine Learning and Fuzzy Logic. Our attempt, finally, results in a useful fuzzy system, which can be used in order to classify firms into efficient and inefficient ones.