| The GTJA-Allianz financial data synthesis management platform and risk control system is an important project of the GTJA Allianz Fund management Limited Company for the adaptation of the current development of technologies and the improvement corporate governance standards and achieving risk control computerized management.The system uses Struts framework and Hibernate persistence layer technology Oracle9i database server background. The software system design top-down roughly four layers: the user interface, expression logic layer, business logic layer, database connectivity layer (previously called persistence layer).This paper focuses on the introducing several functions of the system which are composed by report function, data base maintenance function and risk prediction function. When processing the information of stock, we use the method of association rules to get the results of one-dimensional relationship rules and the inter-time series association rules.The system achives report function using the JasperReport and iReport tools. Among them, JasperReport is one of the report printing components.The report function in system is composed of two phases: the first phase report function and the second report function. Before the realization of first phase report function, we export only the tables on JSP pages into Excel. But the titles of JSP pages and the annotations in the data table are leaved without processing. When the report function is finished, the data such as the titles, the annotations in the data table, etc. will be exported to Excel as well. But the function can't set the width of every column according to appropriate proportion because of some compatibility requirements. So every column has the same width. We use procedures to design the report template according to report parameters. We can use programs to process the report template, and then use the API provide by JasperReport to compile the result file produces by the programs. Although the system is very flexible, the effiency is not very high compared with the second phase report because of the procedure of compiling. After this function is applied the users only need to do some simple editing and then have the report printed. The second phase report is composed of several concrete reports. We use the iReport tool to design several report template and then use Jasper object to realize the report functions. So the efficiency is higher than the first phase. In this paper we will use some examples to introduce the realization of general reports, grouped reports, subreports.The basic data maintenance function of the system is to administrator the several basic data tables in the database. This functions are usually used by the staff who are not familiar with the operation in the database. This function make the staff work more efficiently because it can administrate data tables via simple operations instead of professional implementation in the database.For example, the users can execute query,insert,modify,delete of records in the table AssertPrice.In addition,the function of risk prediction includes two subfunctions: correlativity function and the check of stock pool. The first subfunction calculates the correlativity between the rate of stocks and stocks,bonds and bonds,as well as stocks and bonds.And we present the line chart of the rate of return,which shows the information about the rate of return more directly.And the second subfunction presents details about different stock exchanges between two days.And this make the change of stock exchange more clear to the users.In numerous data-mining methods, association rules,which is simple but practical,has become a hot topic in the field of data mining.Its first successful application is the analysis about the data of market basket transactions. To understand association rules,we should first understand some terminologies and definitions, such as item,transaction,itemset,dimension of itemset,support,confidence. The Apriori is the most famous algorithm for problems like this, and there are several other algorithms which are improved according to the practical demand,such as AprioriTid.The method of association rules is also suitable for the stock datum handling,so we can use it to mine the information in stock datum together with suited concrete algorithm.When mining information in one-dimensional association rules,such as"when the stock of IBM and SUM begin to rise, the stock of Microsoft will rise at the same time by 80% chance.", we can not only set time to start and end,minimum increase,minimum support,minimum confidence and also introduce the concept of sliding window which makes the same period can be one or several days.And the increase is the cumulate increase in these days.After setting the parameters above, we can use the improved algorithm Optimized Apriori introdced in this paper to mine data.The new algorithm put forward a new data struct: which is superior to typical algorithm Apriori in efficiency.And the paper give some explanations about association rules.Besides one-dimensional association rules,we handle multi-dimensional ones and using the improved algorithm HS-Apriori we get the following rule:"when the stock of IBM rises at the first day and the stock of SUM rises at the following day, in most cases the stock of Microsoft will rise at the third day.". HS-Apriori is a extensable algorithm with several stages, and it is suitable for the data mining of inter-time series association rules.In practical applications we get several meaningful association rules.In addition, we also introduce the data mining on the stock information, such as the discovery of period, tense relations, the similarity, the information of users and so on.The system has run for a year in the user's company, many functions are frequently used by customers every day. There is no problem in its correctness and performance. However,to develop a pretty perfect system,there are still some functions to be improved,such as the Report and the application of data mining. |
PDF Full Text Request