| With the arrival of the information and internet age,the world is undergoing unprece-dented change.In the financial sector,big data and artificial intelligence have become the predominant technologies.In the face of massive amounts of investor user data,finan-cial enterprises must undergo digital transformation to meet market demands.This article focuses on analyzing label data for investor users in the financial sector and constructing a user portrait system.It also proposes a cross-feature combination population diffusion algorithm.Additionally,the paper also have designed and implemented a user portrait system based on clustering and diffusion with the aim of providing accurate marketing services for digital transformation of financial institutions,such as securities companies.The research has two specific areas:(1)The proposal of a cross-feature combination population diffusion algorithm that uses EM algorithm and Min-Max data standardization method to preprocess label data for subsequent calculations.PCA,or principal component analysis,is employed to diminish the dimensionality of user labels,thereby enhancing the precision of predictions.Finally,the naive Bayes algorithm is applied to train the seed user label data to identify users with similar features.(2)The design and implementation of a user portrait system based on clustering and diffusion that includes three major business functional modules: label module,user por-trait module,and customer group portrait module.The system also contains two core functions of customer group clustering analysis and population diffusion.The system aims to provide support for the digital transformation of financial institutions,such as se-curities companies,and to provide accurate marketing services to gain more commercial value. |
