Implementation of Genetic Algorithms in the RapidMiner Software

This thesis deals with extending the RapidMiner software with a new operator providing the capability to generate classification rules using a genetic algorithm. The specific system implemented in this case is GAssist (Genetic clASSIfier sySTem). First, an introduction to the topic is presented, including genetic algorithms and the RapidMiner software. Subsequently, the basic theory concerning artificial intelligence and machine learning is described. Classification and genetic algorithms are additionally included and the intersection of those two areas is explored. Following is an overview of the RapidMiner software, focusing on the possibilities of its extension via plugins. One of those approaches, RapidMiner Extension Template, is given special attention as this is the approach used later in the implementation itself. The next chapter, which deals with this implementation, details the structure of the developed plugin, including important classes, algorithms and code snippets. Finally, the implementation is compared with the original GAssist system as well as selected classification algorithms, which are natively supported by the RapidMiner software. After that, the comparisons are evaluated and the results discussed, including differences between both implementations of the GAssist system. The output of this thesis is a plugin for the RapidMiner software implementing the genetics-based classifier GAssist.