In this paper, a hybrid fuzzy clustering techniques using Salp Swarm Algorithm (SSA) is proposed. The proposed fuzzy clustering method is used to optimize the cluster centroids obtained as an under sampling method. The performance of the proposed fuzzy clustering method is compared with some well-k nown clustering algorithms to shows the superiority of the proposed clustering algorithm. In addition, a novel hybrid Automobile Insurance Fraud Detection System is proposed in which undersampling of the majority class is performed by using the proposed fuzzy clustering algorithm which eliminates the outliers from the majority class samples. The balanced dataset for automobile fraud detection obtained after undersampling undergoes classification. Different classifiers used for this purpose are Random Forest Classifier, Logistic Regression Classifier and XGBoost Classifier. The performance of each of the three classifiers is evaluated by considering different performance metrics such as sensitivity, accuracy and specificity. The proposed fuzzy clustering method along with XGBoost outperforms the other methods presented.
K-Means is a popular cluster analysis method which aims to partition a number of data points into K clusters. It has been successfully applied to a number of problems. However, the efficiency of K-Means depends on its initialization of cluster centers. Different swarm intelligence techniques are applied to clustering problem for enhancing the performance. In this work a hybrid clustering approach based on K-means and Ant Lion Optimization has been considered for optimal cluster analysis. Ant Lion Optimization (ALO) is a stochastic global optimization model. The performance of the proposed algorithm is compared against the performance of K-Means, KMeans-PSO, KMeans-FA, DBSCAN and Revised DBSCAN clustering methods based on different performance metrics. Experimentation is performed on eight datasets, for which the statistical analysis is carried out. The obtained results indicate that the hybrid of K-Means and Ant Lion Optimization method performs preferably better than the other three algorithms in terms of sum of intracluster distances and F-measure.
'%3e%3ccircle r='20' fill='%23008'/%3e%3ccircle r='17.5' fill='%23fff'/%3e%3ccircle r='3.5' fill='%23008'/%3e%3cg id='d'%3e%3cg id='c'%3e%3cg id='b'%3e%3cg id='a' fill='%23008'%3e%3ccircle r='.875' transform='rotate(7.5 -8.75 133.5)'/%3e%3cpath d='M0 17.5.6 7 0 2l-.6 5L0 17.5z'/%3e%3c/g%3e%3cuse xlink:href='%23a' transform='rotate(15)'/%3e%3c/g%3e%3cuse xlink:href='%23b' transform='rotate(30)'/%3e%3c/g%3e%3cuse xlink:href='%23c' transform='rotate(60)'/%3e%3c/g%3e%3cuse xlink:href='%23d' transform='rotate(120)'/%3e%3cuse xlink:href='%23d' transform='rotate(-120)'/%3e%3c/g%3e%3c/svg%3e)
