Research And Application Of Financial Big Data Based On Density Peak Clustering Of K Near Neighbors

With the development of technology,data analysis is more and more complex,difficult to process,difficult to classify,and the data dimension is getting larger and larger.Clustering by fast search and find of density peaks(DPC)does not adapt to high dimensionality.The defect of the data set proposes an optimization algorithm called T-DPC,which is based on the t-SNE dimensionality reduction method,and also optimizes the Gaussian kernel function calculation method,and uses uniform metrics when solving the density.In the experiment,the artificial data set and the UCI standard data set were selected.The DPC algorithm was compared with the UCI data set and the T-DPC algorithm.The final experimental results show that the T-DPC algorithm not only adapts to the high-dimensional data set,but also Improve the efficiency of the DPC algorithm.The improved K-nearest density peak clustering algorithm is based on the density peak clustering algorithm(DPC algorithm)and combined with K-nearest neighbor technology to improve the K-nearest neighbor density clustering algorithm.The problem in the original algorithm is solved by improving the adaptive measurement method and the allocation strategy combining K nearest neighbors.Solved the problem of subjective hidden dangers and poor distribution of dc selection.The hottest topic in today's stock market is how to get the most out of it and minimize the risk of earnings.This has become a major research issue for many researchers.The stock market is a complex system,a variety of types,a structural disorder,and various factors affect each other.How to choose stocks is the most concerned issue for investors today.In order to verify the improved K-nearest density peak clustering algorithm applied in financial big data,the experiment selected the stock data of the 20 listed companies in the second quarter of 2018 in the Shanghai and Shenzhen A shares as the financial data,the stock data respectively The data of net profit,earnings per share,return on net assets,operating income,and net assets per share were selected for data analysis.These five data can reflect the comprehensive profitability of the listed company,after data reduction and processing,and then pass k.The density peak clustering algorithm of the neighbors classifies the stocks and selects the blue and the poor stocks.