9 Simultaneous Continuous Feature Selection and K Clustering by Multi Objective Genetic Algorithm

We can classify clustering into two categories. In K Clustering, we know the number of clusters or K. In other category of clustering, K in unknown. In this paper we have considered the first category only. We can broadly classify features within a data set into continuous and categorical. Here we have considered data set with continuous features only. Clustering can be done by all features or by relevant features only. Researches had commonly used some feature selection techniques to select relevant features for clustering and then did clustering by some clustering algorithm. Here we have used Multi Objective Genetic Algorithm (MOGA) for simultaneous feature selection and clustering. Here, K-means is hybridized with GA. We have used hybridized GA to combine global searching abilities of GA with local searching abilities of K-means. Considering context sensitivity, we have used a special crossover operator called "pairwise crossover" and "substitution". Elimination of redundant, irrelevant features increases clustering performance, reflected in MOGA_Feature_Selection (H, S) compared with MOGA (H, S). The main contribution of this paper is simultaneous dimensionality reduction and optimization of objectives using MOGA.