The Research Of Financial Big Data Analysis Service Model And Application

With the continuous increase of the amount of data in financial field,providing the service with high value and high time efficiency for consumers still faces two challenges.One challenge is how to mine the effective value from big data fully and the other is how to deal with big data tasks quickly.Meanwhile,the solution of the above problems will be in-depth study.The arrival of the big data era puts forward new requirements to the processing model and technique of the financial analysis service,and the further study of service model,combination method of analysis services and technology of processing big data to improve service value and time efficiency is necessary.Therefore,the paper first designs a financial analysis service model,the processing flow and system framework,and then researches the combination method of analysis services which is based on multi-view learning and the task scheduling technique,also finally implements a financial analysis service system.The primary work is as follows:1.The financial analysis service model is constructed.The model aims to provide accurate and timely service and it contains four levels designed by hierarchical thought.Based on the study of analytics-as-a-service,the function of each layer is defined formally and the direction for further study is found.In addition,the business process and the system architecture are optimized,and the key technologies to support the architecture implementation are analyzed.2.The method study of improving service value is concerned,and the main idea is utilized multi-view learning method to combine the analysis services.Due to the traditional analysis service is hard to process big data with low density and high dimensionality,the paper proposes a multi-view linear discriminant analysis algorithm.The main principle is to mining the relationship between the data so as to combine multiple atomic services,and then form the specific perspective to train a classifier jointly which can evaluate the effectiveness of the service combination.The experiment shows that the accuracy of the analysis results of multi-view learning is higher than that single-view learning.3.The key technology of optimizing service timeliness is studied,the principal content is the task scheduling technology for heterogeneous distributed computing systems.To meet the scalability needs in the processing technology of big data analysis service,the paper presents a hierarchic hybrid heuristic-genetic scheduling algorithm.It assigned tasks on all processors in reasonable,which can both support parallel computing and satisfy the task dependencies strictly.The comparison performance results show that the algorithm decreases the computational time of multi-view learning tasks and minimizes the execution time of all tasks.4.In the financial securities domain,the application of big data analysis service is achieved.By realizing the service system,it is verified the feasibility of the above three aspects of research from the practice level.