Zühtüoğulları, KürşatAllahverdi, NovruzArıkan, N.2020-03-262020-03-2620131349-4198https://hdl.handle.net/20.500.12395/30156The information based systems with large input spaces require high processing times and memory when soft computing methods are used. Attribute reduction mechanisms are very important these problems by representing the data with the significant attributes. In this study, the genetic algorithm and rough sets based software is designed to realize feature reduction mechanism for the medical databases. The number of the input attributes can be reduced by the GARSBS and the output of the reduction mechanism is combined with variable input neural network classification software. Successful and faster reducing software is developed and a new modified selection mechanism consisting of variable number of generations in the genepool is constructed for obtaining performance. Extreme information storage, time consumption and input number restriction problems of the feature reduction algorithms are solved and classification processing times are reduced by using the generated approach. In the study, urological measurements are accepted as the input variables of the system. The number of the input variables (twenty) has been reduced to the reducts with eight, seven, six and five elements and the classification accuracy has been tested by the artificial neural network part of the software. © 2013 ICIC International.eninfo:eu-repo/semantics/closedAccessAttribute reduction mechanismHybrid rough sets and genetic algorithm based approachRough sets theoryUrological measurementsArticle9730153037Q3