K-means Research And Improvement Based On Particle Group Technology

With the development of With the development of network and information technology,the amount of data is showing an exponential growth trend,and how to obtain useful information from huge data sets is an important research content of data mining at present.Clustering analysis uses a large data set to be divided into multiple sub-categories,so that the data in each sub-category is as similar as possible,and the data between each subcategory is as different as possible,and then the knowledge patterns hidden in the data are explored.The method is an important method used in data mining at the moment.Among them,the K-means algorithm is often adopted by people because of its simple and fast advantages,but the K-means algorithm also has the disadvantages of being sensitive to the initial value and easily affected by outliers on the clustering results.This thesis proposes a CVWK-means algorithm based on cluster error weighting,which improves the clustering effect by weighting clusters with large errors.The specific research work is as follows:Aiming at the influence of each attribute in the data on the final clustering result,a Kmeans algorithm based on cluster error weighting is proposed.The clustering effect is better by adding multiplication to clusters with large errors.After verification on the UCI data set,the improved algorithm is better than the original algorithm.This thesis also combines the proposed K-means algorithm based on cluster error weighting with particle swarm algorithm,and proposes a particle swarm hybrid clustering algorithm based on K-Means algorithm.This algorithm combines the global optimization capability of particle swarms,and takes the obtained global optimal solution as the initial clustering center,thereby optimizing the problem of K-means algorithm.The final experimental results show that the algorithm has a significant improvement over other algorithms in terms of algorithm efficiency and effectiveness.