Modified Approach to Problems of Associative Rules Processing based on Genetic Search
Victoria V. Bova, Sergey Shcheglov, Dmitry Leshchanov · 2019 International Russian Automation Conference (RusAutoCon) · 2019
The paper presents the main problems of associative classification in terms of the processing unstructured data of large scope. There is a review of the popular algorithms for searching associative rules and patterns. The authors formulated the problem of developing a modified approach to the optimization of searching associative rules on the basis of integrating the methods of genetic search and algorithms for associative classification. The solution of the task of searching associative rules includes finding the regularities between the related events in the datasets. The proposed approach allows us to perform the preliminary selection and the further partition of the data into the related groups. The approach also provides the determination of effective methods of the data processing in terms of the Big Data mining. To improve the computational efficiency and the accuracy of solving the task of associative classification, the paper proposes a modified genetic algorithm. The experimental research of the effectiveness of the developed algorithm are based on the task of optimization of the input data filtering in terms of the benchmark dataset Retai.