Research On Multi-source Forecasting Information Fusion Of Security Alalysts Groups Based On Two-Dimensional Evidence Theory

Work of securities analysts is closely related to the development of the capital market.Securities analysts pass on professional information in the form of research reports,which further narrows the information asymmetry between investors and investees.From the perspective of the practice of domestic and foreign capital market,work of securities analysts play an important role in screening the quality of listed companies,improving market efficiency and allocating resources rationally.The development of internet,big data and artificial intelligence technology makes it possible for people to process and utilize the research reports of securities analysts.Integrating the decision-making information of exports group has become the frontier of expert group intelligence research,but due to the differences in individual securities analysts,the limitation of their own knowledge and understanding of the problem,securities analysts will give prediction information with different importance and reliability,and this part of information has an important impact on the quality of the final synthesis result of the prediction information.Therefore,how to measure the reliability and importance of prediction information given by securities analysts and modify the original prediction information to effectively use the prediction information is an important scientific problem that must be solved when solving complex decision-making problems with the knowledge and experience of the securities analysts groups.This article introduces the two-dimensional evidence theory into the research of the multi-source prediction information integration model of securities analysts as the basis theory to solve the above-mentioned problems,and discusses the generation of evidence reliability function and the synthesis of evidence in different dimensions based on two-dimensional evidence inference rules.For the problems to be solved,this paper uses data exploration,two-dimensional linguistic assessment information,gray theory,performance study and other methods to further optimize the theory of two-dimensional evidence synthesis method,focus on solving the subjective information evaluation problem of language,the measurement method of securities analysts ability index and the synthetic optimization problem of two-dimensional evidence theory,builds a multi-source prediction information integration model of securities analysts based on two-dimensional evidence theory for the multi-source prediction information integration of securities analysts groups.(1)It provides a general analysis framework for multi-source prediction information integration of securities analysts groups.Multi-source prediction information integration of securities analysts groups is a typical uncertain decision-making problem.Evidence theory has obvious advantages in solving this kind of problems,but traditional evidence theory fails to consider the difference of the quality and reliability of information from different information sources,and fails to reflect the relevant process and characteristics of securities analysts' decision-making conclusions based on historical data(empirical evidence)and knowledge experience and the reliability functions obtained from different information sources.This paper explores classical theory of evidence identification framework and systematically integrates personal characteristic information of securities analysts to form two-dimensional evidence identification framework that can fully reflect the quality of prediction information of securities analysts,forms a multi-source prediction information integration model of securities analysts based on two dimensional evidence theory and further improves the synthesis rules of the classical evidence theory.(2)It comprehensively reflects the analysis and prediction ability of securities analysts from multiple dimensions and takes it as an important factor of prediction information integration.Among the existing information integration methods of securities analysts,most of them directly use the objective stock recommendation information given by the security analyst and fail to fully use the personal characteristic information of the security analyst.This paper comprehensively reflects the future prediction ability of securities analyst by systematically sorting out existing relevant theoretical literatures,referring to the classic ability framework of the financial analyst and forming the comprehensive index of the comprehensive prediction ability of the securities analyst based on the work experience,academic background,job performance and research report quality of the securities analyst.While helping decision makers to reasonably select decision-making information,it also provides brokers with a new idea to evaluate the comprehensive prediction ability of the securities analyst.(3)It deeply analyzes the subject and object information of the securities analyst's research report.Most of the existing researches only focus on the final stock recommendation information and key index prediction and other quantitative information in the research report of the security analyst,and fail to fully use the qualitative evaluation information in the research report of the security analyst,while the above-mentioned information is the key information that reflects the quality and characteristic of the research report.By systematically sorting out relevant literatures and counting the content of the research reports of a large number of securities analysts,this paper proposes that the frequency of occurrence of core keywords in research reports can reflect the familiarity of securities analysts and the quality of research reports to some extent.Therefore,security analysts' ability includes not only the qualitative indicators such as the personal characteristics of securities analysts,but also quantitative indicators such as the quality of research reports.This paper organically integrates the subjective and objective weighting methods,and uses the grey theory and whitening weight function to scientifically and effectively integrate qualitative and quantitative data to complete the consistency transformation of mixed indicators.(4)It proposes a method that comprehensively optimizes two-dimensional evidence integration internal structural parameters based on learning.Two-dimensional evidence needs to fit the second-dimensional evidence information into the traditional first-dimensional evidence framework in the process of information synthesis.This paper fully uses the historical prediction data of securities analysts and refers to the method of supervised learning,learns and optimizes the internal parameters of the two-dimensional evidence system,and focuses on the multi-source prediction information integration of securities analysts,constructs the performance matrix through the prediction accuracy of securities analysts,so as to make the obtained comprehensive optimization parameters have dynamic adaptability.(5)Based on the experimental study of multi-source prediction information integration model of securities analysts groups based on the two-dimensional evidence theory,this paper systematically verifies the practicability and effectiveness of the proposed method.In the early research to obtain the real data of Anhui Guoyuan Securities security market,the research studies the multi-source prediction information method of securities analysts groups based on the two-dimensional evidence theory.Taking the final prediction accuracy of the securities analysts groups as the evaluation standard,by comparing the traditional methods of information integration based on statistics,this paper explains the rationality of the comprehensive prediction ability of securities analysts,and verifies the practicability and effectiveness of the application of two-dimensional evidence theory in the process of multi-source prediction information integration of securities analysts.