A Study And Practice Of Data Analysis System Based On The Technology Framework Of SOA

With the development of Information Technology, Data Mining, Data Analysis and other related technologies develop rapidly. Data Mining and Analysis can turn historical data tombs into auxiliary decision-making golden nuggets of knowledge, which can give effective help to decision-makers."Automatic Fare Collection (AFC) operation management data analysis system of rail transit", which is currently suitable for the industry characteristics of AFC in the country, is a data analysis application software system with the general and special statistical analysis. This paper, carried out for further study based on the project above, design and practise the data analysis prototype system for Rail Transit AFC operation and management, based on the technology framework of SOA, study and implement an improved local outlier mining algorithm, and apply the improved algorithm to the data analysis of AFC.The primary research works in this thesis are described as followings:1) Study the implementation of Web Service based on the technology framework of SOA, and apply it to the data analysis system, which is useful to enhance the reusability of software modules and system scalability.2) Study the architecture pattern of AFC operation management data analysis system of rail transit, design and implement the Five-Layers Architecture Pattern data analysis prototype system for Rail Transit AFC operation and management based on SOA technology framework, which is the extension of Three-Layers Architecture Pattern based on design patterns.3) Study outlier mining algorithms, introduce and implement an improved algorithm for local outliers mining based on cluster analysis, and integrate it into the data analysis prototype system for Rail Transit AFC operation and management based on SOA technology framework in the form of Web Service.4) Apply the improved algorithm for local outliers mining based on cluster analysis to the data analysis prototype system for AFC operation and management, show the application effectiveness of the improved algorithm above.In this paper, a data analysis prototype system is designed and implemented, based on SOA technology framework. The prototype system is much easier to be extended and maintained, and besides, the business modules in the form of Web Service have much better reusability. The improved algorithm for local outliers mining based on cluster analysis has much better performance, shows the superiority with respect to the effect of mining, and has a certain academic value.