User Behavior Analysis Algorithm And Its Application On Spark

With the rapid development of Internet,big data has become one of the hot topics of the times.In this context,the amount of users' behavior data is increasing rapidly,and the business needs to understand the users' interests by analyzing these data so as to achieve the goal of maximizing profit by making accurate advertising and personalized recommendation to the users what they need,and the users also need business provides targeted information that they are interested for them after analyzing their own behavior data,so as to enjoy more convenient and efficient services.At present,many scholars have done some researches on user behavior analysis.However,with the increasing amount of data,the traditional database technology has been difficult to meet the needs.In recent years,a variety of large data technology came into being.Hadoop,Spark and other platforms provide technical support for dealing with massive user behavior data.After studying various learning methods of neural network,aiming at the shortcomings that the convergence speed of neural network based on gradient descent algorithm is slow and easy to fall into local optimum,and the computing of neural network based on the evolutionary algorithm is too much,this paper proposed a bird swarm algorithm-neural network,which uses bird swarm algorithm to improve the neural network.The algorithm uses the characteristics of the swarm intelligence algorithm,whose simple calculating to speed up the learning speed of the neural network,and the strong global search capability to improve the accuracy of the neural network.At the same time,the problem of falling into the local optimum is avoided by using the concept of migration in the BSA algorithm.On this basis,this paper analyzes the user behavior data of the e-commerce website,establishes the corresponding neural network model and extracts the input characteristics of the neural network to predict the consumer behavior.Through training and testing,the effect of BSANN algorithm and BP algorithm is compared,and the feasibility and effectiveness of the BSANN algorithm are verified.Based on the above research,this paper implements the parallel BSANN algorithm on the Spark platform,designs and implements a user behavior prediction system based on Spark.This system takes the BSANN algorithm which is based on bird swarm algorithm as the core to predict user consumption behavior of e-commerce website.