Research On The Recommendation Algorithm Of Financial Products Based On Machine Learning

Nowadays,the recommendation system occupies a pivotal position in both e-commerce and social networks.The traditional financial institutions represented by banks in the application of personalized recommendations are slightly less than the Internet industry.This paper uses the latent factor model based on machine learning method and combines the characteristics of users to generate an improved latent factor model of user characteristics,so as to recommend financial products.When solving the cold start problem,the new user can be recommended by the user's demographics.The method calculates the user's characteristics about the interest of the item,and accumulates all the characteristics about the interest of the item,thereby obtaining the user's degree of interest of the item.The latent factor model calculates the user's interest_,6)for the class and the item's weight₄,6))for the class by classifying the items.Then multiply the two to get the user's interest in the item.The combination of the two recommended methods is the user's interest in the item.In this paper,the user's interest degree based on the user's characteristics is linearly combined with the interest calculated by LFM,and the weight of the two is measured by the user's activity.Use the trained model to recommend the topitems with high target interest?TopN?.Through the experimental test on the Movielens dataset and bank wealth management product dataset,the improved latent factor model with user features is better in recommending performance when the sample size is small and the user cold start problem is dealt with.