Cluster generator for cluster analysis

The aim of this diploma thesis is to extend the desktop application of the cluster generator created within the bachelor's thesis "Design and development of a cluster generator for cluster analysis". This work summarizes the issue of generating random numbers according to a certain probability distribution captured in the bachelor's thesis and the development of this application. The Java programming language was used for its development. The first part briefly summarizes the statistical methods and theories that occur in the application and are needed to understand the issue. Subsequently, in the second part, the requirements for the development of this application are processed and described. All these requirements were created on the basis of questionnaires provided to people who used the application and had ideas for its improvement. The final part deals with the implementation of defined requirements. The conclusion of the thesis contains a summary of the achieved results and an evaluation of the whole work. The result of this work is a functional application for generating clusters using numbers generated using a specified probability distribution. This can be used mainly in cluster analysis, but also in discriminant analysis, which is another multidimensional statistical method. Outputs can be used for this method because the user knows the division of objects into groups in advance, which is a condition of this method. When creating clusters, the user can select the number of clusters created, the number of variables that characterize each object, the type of probability distribution, variability within clusters, the degree of cluster separation given by their cluster overlap percentage, and other modifiable end-user inputs. Based on this modified application, it will be possible to use and evaluate clustering methods from various perspectives, which still creates space for further scientific research, including the already mentioned discriminant analysis.