An Improved K-means Clustering Algorithm Based On Artificial Bee Colony Algorithm And Its Application

Artificial Bee Colony Algorithm(ABC)is a group of intelligent optimization algorithm for simulating the cooperative behavior of bees,which was systematically presented by the Turkish scholar Karaboga in 2005.Because of the characteristics of simple operation,less control parameters,faster convergence,easy to achieve and so on,it attracts more and more scholars to study and effectively applied to function optimization,data mining,neural networks,travel agents and other practical optimization problems.However,in the process of optimization of the basic artificial bee colony algorithm,the optimal solution is chosen by the greedy method which results in the insufficient search of breadth and depth,the lower search accuracy,easy to fall into the local optimal solution,and the impact of algorithm performance,thus,many scholars have proposed varied mathods to improve the basic ABC algorithm.The main works of this paper are:(1)Because of the Gbest-guided artificial bee colony algorithm(Gbest-guided ABC algorithm)proposed by Zhu Guopu et al,does not fully consider the effect of global optimization and local optimization in the optimization process of the search iteration process,the global search ability of the algorithm is easy to fall into the problem of local optimal solution.In this paper,a global optimal artificial bee colony algorithm(HF-GABC)with hunting factor is proposed.On the basis of GABC algorithm,we introduce the dynamic search function with the optimization process,and carry on the dynamic search to the global search process and the local search process.The improved algorithm is used to test on the four standard test set functions and compared with the ABC algorithm and GABC algorithm.The experimental results show that the convergence performance of the artificial bee colony algorithm with hunting factor is better than ABC and GABC algorithm,which effectively reduces the possibility of local convergence and improves the search precision.(2)Because the traditional K-means clustering algorithm has the fast convergence speed but it is sensitive to the initial clustering center,easy to fall into the local optimum and weak robustness,a clustering algorithm combining the global optimal artificial colony algorithm with hunting factor(HF-GABC)and K-means algorithm is proposed in this paper.The improved global optimal artificial bee colony algorithm can improve the local search ability and global search ability,and combine with the advantages of fast convergence with k-means clustering algorithm to solve the problem that the original K-means algorithm is too dependent on the initial clustering center and easy to fall into the local optimal defects.In order to verify the feasibility and effectiveness of the algorithm,we choosed several data sets in the UCI machine learning database for the experiments.The experimental results show that the new algorithm not only overcomes the shortcomings of the traditional k-means robustness,but also improves the clustering effect.(3)The improved clustering algorithm was applied for the e-commerce customer's classify of a website's transaction data.The steps of the improved clustering algorithm in the customer's classify are discussed in detail.The K-means and HFGABC-K are used in the same group of data to segment the work and the results were analyzed.The results show that the clustering effect of HFGABC-K algorithm is more compact and excellent,and the quality of classfiy is better than that of K-means.According to the results of clustering evaluation,the algorithms can be usedto help enterprises to develop appropriate marketing strategies.