Personalized Recommendation Algorithm Based On Multivariate Statistics

With the rapid development of the Internet,the exponential growth of information presents users with information overload,making it difficult for users to get the information they want.In this case,personalized recommendation technology comes into being.Personalized recommendation technology is mainly used to allocate resources efficiently and reasonably so that users can quickly get the information they want from a large amount of information.As a collaborative filtering recommendation technology widely used in personalized recommendation technology,although it has achieved great success in practical application,it still faces many technical difficulties.This paper introduces the system of theoretical knowledge and the classical recommendation algorithm,and then focuses on the principal component analysis and K-means the combination of clustering algorithm and collaborative filtering algorithm,Finally,the effectiveness of the improved algorithm is illustrated by experimental analysis.The main work of this paper is as follows:1.Starting from the time effect of user behavior data,this paper studies the recommendation algorithm.The analysis of users' interest is not the same,but changes over time.Aiming at this deficiency,the time forgetting function improvement algorithm is introduced.At the same time,the influence of hot items on user similarity calculation is analyzed,and the penalty factor is introduced to reduce the influence of hot items on user similarity.Finally,the experiment shows that the improved algorithm with time forgetting function and penalty factor is more effective in predicting scoring accuracy.2.In view of the traditional collaborative filtering algorithm to achieve the data calculated result in memory consumption and time of the bottleneck problem,was proposed based on PCA and K means clustering algorithm hybrid recommendation algorithm,based on user ratings matrix dimension reduction,greatly reduces the computational complexity.Finally,the improved algorithm is more effective when adding principal component analysis and k-mean clustering algorithm to improve the accuracy,recall rate and comprehensive evaluation index.