Bank User Behavior Analysis Based On Distributed Stored Database

Since the big data strategy was put forward by the state,the Internet ushered in the spring.With the increase of the registered volume and activity of users,the amount of data has also increased rapidly.With the rapid growth of information,the increment of traffic and financial consumption data,daily consumption of financial log and transaction records reaches the TB level,the traditional data storage capacity and access speed is inadequate for the rapid growth of the amount of data,which leads to information overload problems.At present,there are two ways to deal with the problem of information overload,one is to increase the server to increase storage,and the other is to partition the data storage index.But with the increase in the amount of data it has other issues,the first increasing in server will increase service cost,then establishing the index can't retrieve data fast in the massive data scene,which can't quickly and accurately mining users' behavior from massive data.The development of the Internet technology brings the advent of big data era,big data analysis platform extracts the user's behavior quickly and accurately from the massive data,which is very important for precision marketing,improving system performance and service.This article starts with the collection of information,and uses JavaScript technology to crawl user's behavior data on the loan application website.We compute the data grabbed by real-time computing called Storm,on the one hand the results are stored in the memory database called Redis,and on the other hand,the results will be stored in the distributed database HBase.In order to solve the problem of query response efficiency,this paper uses the Solr index technology on the basis of large data platform to make up for the problem that HBase cannot achieve the combination condition query.This research project builds up a user behavior analysis platform through a large amount of literature and application experience of enterprise users on the big data structure of Hadoop.The main research contents include:(1)Research and development of data acquisition system based on JS data capture technology;(2)Research and development of distributed storage database based on Hadoop offline computing and real-time flow computing Storm which are distributed computing framework;(3)to solve the problem that the HBase can't query by combination query condition by studying HBase combined with Solr;(4)to study on using clustering method of statistical user behavior;(5)to study the Rowkey and optimize the HBase parameters to improve the performance of the database.