Enterprise Business Analysis And Prediction Based On Machine Learning And Knowledge Graph

Enterprise development is an important part of social and economic development,the government should monitor the development of enterprises every year,timely adjustment of the list of cultivation enterprises,how to efficiently select and develop leading enterprises is a difficult task.All kinds of data resources as a whole,the analysis and prediction of construction enterprise management system,for local governments to provide related services,enterprise development is not only beneficial to optimizing the allocation of service resources,more conducive to speed up the industrialization and information fusion,play the advantages of emerging enterprises,to regional construction level and meet the demand for the diversification of information service in the process of enterprise development has the vital significance.However,the current research on enterprise operation and development forecast mainly focuses on the use of single data to forecast the development of enterprises,or the research uses time,weather and other factors,without considering the sudden situation in reality,but can only predict the development of enterprises under regular conditions,unable to respond quickly to the impact of irregular factors,affecting the accuracy of the forecast.In addition,considering the impact of various factors on the business situation of enterprises,there will be data privacy problems encountered in the data collection,how to apply the forecast to the practice after the completion of the forecast,using the data of multiple enterprises to make a more effective assessment of the industry development and other issues,there are still many challenges in the actual business forecast.Aiming at the above problems,this paper uses machine learning and knowledge graph to analyze and predict enterprise operations,which can solve the problem of data privacy security on the basis of ensuring the accuracy of prediction results.And compare enterprises according to industry classification to provide better data support for decision making.The main work of this paper is as follows:(1)Prepare data for business analysis and forecasting.According to the requirements of the system,this paper analyzes the data characteristic dimensions needed for prediction,and puts forward the non-quantitative data knowledge representation method based on knowledge graph according to the data characteristics,quantifies the required data,and completes the data preparation required by the enterprise management prediction model.(2)The enterprise operation prediction algorithm based on machine learning and federated learning is designed.The model is trained according to the characteristics of the collected data without summarizing the data,which not only completes the more accurate enterprise operation prediction,but also protects the data privacy.(3)This paper proposes an industry matching method based on industry constraints,which achieves accurate matching between enterprises and industries without additional data annotation,and has high accuracy.The information between enterprises can be displayed and compared according to the industry,which solves the problem that enterprises in different industries cannot be compared.(4)An enterprise management analysis and forecasting system for government is designed and implemented.The system includes four modules: user management module,data management module,enterprise management analysis module and enterprise management prediction module,which provides multi-dimensional data reference for government personnel to understand enterprise development.