This paper describes the implementation and the functioning of RAGA (Rule Acquisition with a Genetic Algorithm),a genetic-algorithm-based data mining system suitable for both supervised and certain types of unsupervised knowledge extraction from large and possibly noisy databases.The genetic engine is modified through the addition of several methods tuned specifically for the task of association rule discovery.A set of genetic operators and techniques are employed to efficiently search the space of potential rules.During this process,RAGA evolves a default hierarchy of rules,where teh emphasis is placed ont he group rather than each individual rule.Rule sets of this type are kept simple in both individual rule complexity and the total number of rules are kept simple in both individual rule complexity and the total number of rules that are required.In addition,the default hierarchy deals with the problem of over-fitting,particularly in classification tasks.Several data mining experiments using RAGA are described.
展开▼
