Analysis Of Customer's Purchase Of Investment And Finance Products Based On Improved K-means Algorithm

The analysis of customers is very important for all walks of life.It is critical to be able to accurately divide customers into corresponding groups and conduct in-depth analysis.Among them,K-means algorithm is often applied in customer segmentation.It is the advantage of K-means that the operation is simple,the principle is easy to understand,but the clustering result is very easy to be influenced by the initial point,isolated point and noise,and this algorithm needs to offer K value in advance.Therefore,the paper proposes corresponding improvements on its shortcomings.The idea is to use the improved UPGMA algorithm to filter out the candidate center points,and then use the selected candidate center points as the input data of the maximum-minimum distance algorithm,because of the maximum-minimum algorithm.Need to use parameters,so this paper uses the idea of deconvergence to decompose the corresponding interval into multiple smaller intervals,and use the attribute discretization to deal with,and then use BWP indicator function results to evaluate,according to the BWP index by the large and small Take the average of 95% as the optimal number of clusters.The algorithm solves the problems mentioned above with traditional K-means well.In order to verify the effectiveness of the algorithm,Iris,Glass and New-thyroid libraries were selected and analyzed by different poly algorithms.The results show that the improvements in this paper are feasible.The paper selects the customers of e-Baifu,part of the investment banking product of Rural Commercial Bank of Nanchang City,Jiangxi Province,as the research object.In the experiment,the improved algorithm,the traditional K-means algorithm,and the comparison experiment based on the maximum-minimum-distance algorithm are used to improve the experiment.It works.The paper integrates the above-mentioned clustering results to analyze different customer groups,labels the corresponding class names,and then formulates sales plans based on the classification,so as to improve the overall service and product quality of the bank.