Now showing 1 - 6 of 6
  • Publication
    The Concept of Ontology for Numerical Data Clustering
    Classical clustering algorithms have been studied quite well, they are used for the numerical data grouping in similar structures - clusters. Similar objects are placed in the same cluster, different objects – in another cluster. All classical clustering algorithms have common characteristics, their successful choice defines the clustering results. The most important clustering parameters are following: clustering algorithms, metrics, the initial number of clusters, clustering validation criteria. In recent years there is a strong tendency of the possibility to get the rules from clusters. Semantic knowledge is not used in classical clustering algorithms. This creates difficulties in interpreting the results of clustering. Currently, the possibilities to use ontology increase rapidly, that allows to get knowledge of a specific data model. In the frames of this work the ontology concept, prototype development for numerical data clustering, which includes the most important characteristics of clustering performance have been analyzed.
    Scopus© Citations 1
  • Publication
    Many educational courses operate with models that were previously available only in mathematics or other learning disciplines. As a possible solution, there could be the use of package IBM SPSS Statistics and Modeler in realization of different algorithms for IT studies. Series of research were carried out in order to demonstrate the suitability of the IBM SPSS for the purpose of visualization of various simulation models of some data mining disciplines – particularly cluster analysis. Students are very interested in modern data mining methods, such as artificial neural networks, fuzzy logic and clustering. Clustering methods are often undeservedly forgotten, although the implementation of their algorithms is relatively simple and can be implemented even for students. In the research part of the study the modelling capabilities in data mining studies, clustering algorithms and real examples are demonstrated.
  • Publication
    Different Approaches to Clustering – Cassini Ovals
    Classical cluster analysis or clustering is the task of grouping of a set of objects in such a way that objects in the same group are more similar to each other than to those in other groups or clusters. There are many clustering algorithms for solving different tasks. In the research, an interesting method – Cassini oval – has been identified. The ovals of Cassini are defined to be the sets of points in the plane for which the product of the distances to two fixed points is constants. Cassini ovals are named after the astronomer Giovanni Domenico Cassini who studied them in 1680. Cassini elieved that the Sun travelled around the Earth on one of these ovals, with the Earth at one focus of the oval. Other names include Cassinian ovals. A family of military applications of increasing importance is detection of a mobile target intruding into a protected area potentially well suited for this type of application of Cassini style method. The hypothesis is proposed that the Cassini ovals could be used for clustering purposes. The main aim of the research is to ascertain the suitability of Cassini ovals for clustering purposes.
  • Publication
    This paper describes one of classification algorithms, cluster analysis, that plays a significant role in the implementation of learning algorithm as applied to RBF-type artificial mural networks. The mathematical description of the K-means clustering algorithm is given and its implementation is demonstrated by experiment.
  • Publication
    Application of Fuzzy Rule Base Design Method
    In many classification tasks the final goal is usually to determine classes of objects. The final goal of fuzzy clustering is also the distribution of elements with highest membership functions into classes. The key issue is the possibility of extracting fuzzy rules that describe clustering results. The paper develops a method of fuzzy rule base designing for the numerical data, which enables extracting fuzzy rules in the form IFTHEN. To obtain the membership functions, the fuzzy c-means clustering algorithm is employed. The described methodology of fuzzy rule base designing allows one to classify the data. The practical part contains implementation examples.
  • Publication
    Prediction of corporate bankruptcy is a study topic of great interest.Under the conditions of the modern free market, early diagnostics of unfavourabledevelopment trends of company’s activity or bankruptcy becomes a matter ofgreat importance. There is no general method which would allow one to forecastunfavourable consequence with a high confidence degree. This paper focuses onthe analysis of the approaches that can be used to perform an early bankruptcydiagnostics- in previous research multivariate discriminant analysis (MDA), neuralnetwork based approach and rule extraction method have been examined. Lately,time series clustering approach has become popular and its feasibility forbankruptcy data analysis is being investigated. Experiments carried out validatethe use of such methods in the given class of tasks. As a novelty, an attempt toapply time series clustering method to the analysis of bankruptcy data is made.