Application Of BI Data Warehouse And Prediction Technology In Financial Sharing And Analyzing System

Based on the idea of business intelligence and according to requirements of a bank's information construction, this paper implements the financial sharing and analyzing system, using the data warehousing, on-line analytical processing, data mining and other related technologies.Business Intelligence is a new interdisciplinary subject. It brings together multiple achievements from other subjects. And it can realize the effective integration, management and analysis of existing business data and information, making the technology serve to the decision-making reasonably.During the years of financial information construction in our country, we has set up a number of excellent business information systems and accumulated a large amount of valuable data. However, because of the independence of these systems, the large number of valuable information can not be well used. Therefore, this article does some research into related technologies and theoretical approaches of business intelligence and then applies them to the construction of financial sharing and analyzing system, providing a good reference about how to solve the above-mentioned problems. In this paper, there are seven chapters and it can be divided into three parts.The first part contains chapter 1, chapter 2 and chapter 3.In this part, the basic theories and approaches are illustrated. The details read as follow: The current situation of the application of business intelligence technology both here and abroad; Description of the key concepts of data warehousing, multidimensional data model and its implementation patterns, ETL process and its implementation methods, OLAP features and basic operation and so on; discuss the linear regression model and gray model in detail, put forward a prediction model which is based on a combination of distance and regression using gray model to amend the residual error. In this model, we provide the methods of how to get the variables which have a linear relationship and how to identify the sign of the residual error. And we do some experimental analysis of the model.The second part contains chapter 4, chapter 5 and chapter 6.The requirement analysis and the overall designs, each modular implementations and tests of the system are shown in this part. It includes system introduction, as well as functional and non-functional requirement; system architecture, conceptual model and logical model of data warehouse and the implementation of the models, the design and implementation of data integration, data analysis and other functional modules; the application and test of the system.The third part contains chapter 7.In this part we evaluate the system, considering the limitations of the system and the vanguard technology; we give the prospects about application of business intelligence solutions in the coming research work.The financial sharing and analyzing system is composed of several subsystems, they are data collection, data integration, extempore query, multi-dimensional analysis, mining analysis, data sharing and system management. Each functional module also includes a subset of the corresponding functions.In order to make effective integration of the decentralized and unavailable data, data collection and integration are the primary functions to achieve in the system. The data sources of the system are extremely complex, including business systems, reports and common data files which are not only in different format but also distributed in a variety of platforms. So, we design and implement many data collection methods, such as tool extraction, file upload, template data collection and on-line input. And use the Web Service and the Quartz Scheduler to package certain collection mode into an automatically or manually mission, implementing the mission-style collection. When it comes to the file data collection which needs transmission by network or medium, the system uses DES encryption to ensure the security. In the data integration module, we use the transformations provided by INFORMATICA to implement the data conversion. We also develop generic abstract data types and fixed conversion modes for special data.Through data collection and integration, systems build a financial data warehouse, including an item system which contains about 1500 items and their properties and granularities, all statistics about business subjects, the relevant dimension information and user information.We make use of the information in financial data warehouse through the extempore query, multi-dimensional analysis and mining analysis which are provided by the system. Extempore query is not specific query program. It is based on user input, uses the item system to create the query plan, and then shows the results to users by tab cards or graph. In multi-dimensional analysis module, we use COGNOS to finish the multi-dimensional modeling, and implement the OLAP and report analysis by Report Studio and Analysis Studio. In addition, the system also provides early warning analysis and statistical analysis functions. The early warning analysis is implemented by creating specific warning formula of real or derived items; users can monitor the item information or track the detail information about the derived item. In statistical analysis, the system provides a multiple linear regression prediction, gray prediction for single-item time series data, as well as the combination model prediction. The system also provides a good data interface for the third-party analysis tools, users can export data files in various forms.In addition to data integration and analysis, the system provides data-sharing service, supplying specific data to users. The system also provides effective management and maintenance of item system, user information, log, privilege, a variety of mission information and roles to ensure the integrity of the system.In the actual design and development of the system, we use B/S architecture and apply the MVC pattern. The main technologies used are Webwork, Spring, Hibernate and Web Service. Webwork combining the JSP are used for view layer development; Spring is responsible for the management of business entities, service classes and transaction mechanism in the project. Hibernate is mainly used for the development in persistence layer, providing data access. Web Service is for the implementation of specific functions.In this paper, all the proposed elements are specific practical problems. The completion of these elements has solved the problems that the financial data is scattered and difficult to use. Achieve the deep-processing and analysis of information. At the same time, it also provides a successful case for the financial information construction and lays a foundation for further work.