The Improvement Of Seveal Algotithms Based On Density And The Application In The Evaluation Of E-commerce Websites

Clustering analysis is an important technology in data mining, and it is an important method or means of acquiring knowledge. DBSCAN algorithm, as the classical density-based clustering method, can find the high-density clusters with arbitrary shape in the dataset containing noise and the clustering result is not be affected by noise. However, DBSCAN also has the following drawbacks: prior knowledge is required to set parameters manually; it is sensitive to the input parameters, especially when dealing with uneven density data, clustering result obtained is poor. To solve the above-mentioned problem, this paper proposes a DBSCAN algorithm based on data partition. The basic idea of ??the clustering algorithm: firstly, combining AFSA and k-means to realize the dataset partition; then, introducing the coefficient of variation to optimized the parameters Eps on the basis of data partition; finally, running the DBSCAN algorithm on each data partition and getting the final clustering results.In order to verify the effectiveness and usefulness of the improved algorithm, this paper applies the improved DBSCAN to the evaluation of enterprises e-commerce websites. With the rapid spread and development of the Internet and e-commerce, most enterprises have already set up their own e-commerce websites on the Internet in order to promote enterprises' competition. However, lacking of standard evaluation criteria,most enterprises do not have the proper understanding of e-commerce websites.Therefore, making a rational analysis and evaluation of e-commerce websites will help enterprises to improve the construction of e-commerce websites and achieve improvement and self-development. In addition, data of e-commerce websites have uneven density, so it is suitable for the improved clustering algorithm DBSCAN to cluster, analyze and evaluate e-commerce websites. The successful solutions to e-commerce websites evaluation will also help DBSCAN algorithm in the practical applications of a more in-depth study.