K-means Clustering Analysis And Application Based On Evolutionary Firefly Algorithm

As a classical clustering algorithm based on partitioning,K-means clustering algorithm is widely used because of its simple principle,fast convergence speed and easy implementation.However,K-means clustering algorithm is more sensitive to the selection of initial clustering centroids,and different initial clustering centroids often result in different clustering results.At the same time,the rise of group intelligent bionic algorithm provides a new idea for data analysis.The firefly algorithm is inspired by the survival characteristics of firefly populations in the biological world.Firefly algorithm has the advantages of global optimization ability,less input algorithm parameters and easy implementation.However,in the later iteration of the algorithm,it tends to generate strong fluctuations near the optimal solution.Moreover,it is difficult to jump out of the local optimum.In view of the shortcomings of the above firefly algorithm,this paper is inspired by the idea of mixed frog leaping algorithm and introduces the idea of randomized grouping on the firefly algorithm,and then introduces the dynamic inertia coefficient and the moving direction factor to the firefly algorithm.Finally,the combination algorithm is used to perform the detection of outlier,and the application of the time series model is completed.Specific work is as follows:? Firefly algorithm global optimization ability is the most important.In order to ensure and enhance the global optimization performance of the algorithm,this paper introduces the idea of random grouping on the basis of the original algorithm.With the evolution of each iteration of the population for a random grouping,we can achieve the goal of group optimization firstly and then population optimization.? For the displacement formula of the original firefly algorithm,this paper improves it by adding the dynamic inertia factor and moving direction factor,and making the firefly from the individual in the optimization process can better approach the optimal solution,while more ability to jump out of the local optimal and the trend of global optimization.? In the process of combining the improved firefly algorithm with the traditional K-means clustering algorithm,we optimize the optimal solution by choosing each data in the data set to be a firefly individual and the intra-cluster variation is taken as the objective function.Finally,the clustering analysis of K-means clustering algorithm under the condition of initial cluster centroid is finished.? In order to ensure the accuracy of the prediction results of the time series model,we use the above-mentioned combination algorithm to detect the outliers in the dataset and eliminate the outliers,and then use the Newton interpolation algorithm to fill the data.This not only ensures the accuracy of the data set,but also to maintain the integrity of the data set.Finally,the feasibility of predicting the data set after processing in this way is shown by experiments.Experiments on the data set show that the improved firefly algorithm has high stability and high efficiency.At the same time,it is combined with K-means clustering algorithm to obtain high accuracy when clustering the data set.Finally,the feasibility of outlier detection of the data set is shown by the experimental results of the prediction of the monthly CO2 content in the northern area of Canada.