Evolutionary Nearest Neighbour Classification Framework.

摘要: Data classification attempts to assign a category or class label an unknown data object based on available similar set with labels already assigned. K nearest neighbor (KNN) is widely used technique in mining. KNN assigns the majority of its closest neighbours object, when classifying object. The computational efficiency and accuracy depends largely techniques identify neighbours. selection similarity metric optimum as number can be considered optimization problem. optimizing parameters for are value K, weight vector, voting power neighbours, attribute instance selection. Finding these values search problem large space. Genetic Algorithms (GA) provide solutions problems space defined by application domain. There multiple real world applications that utilize parameter optimized KNN. Due this, there various research work carried out using classification. Even though instances had been GAs optimize no software framework available, which could easily adapted domains. This aimed towards building carry help Algorithm. developed provides basic backbone GA while providing sufficient flexibility user, extend it specific