A Company Based On FCM Clustering-Random Forest Algorithm Business Risk Early Warning Research

Risk early warning is an important part of a company’s operation.It is an important way for a company to reduce risks by identifying,evaluating,predicting and preventing risks through the risk early warning system.How to make quantitative analysis of the data and use the analysis results to create a risk early warning model is an important process for a company to achieve digital transformation,cost reduction and efficiency improvement,and industrial upgrading.At present,most of the company’s risk warning only uses the relevant departments to identify and qualitatively analyze the risks that occurred in the past,but fails to use scientific and effective quantitative methods for risk warning,and there is no corresponding process for prevention and control.Therefore,how to use historical data for quantitative analysis and risk identification,evaluation,prediction,prevention and control are the pain points and difficulties of the company.By studying the process of risk early warning,based on the analysis of domestic and foreign research status such as risk early warning and intelligent algorithm,and based on the theories of business risk,risk management and risk early warning,this paper finds the problems existing in the risk early warning of C company through the analysis of the basic situation and early warning status of the research subject C company,and it is necessary to improve the risk early warning system.To build an effective risk warning system for C company.Firstly,risk identification is carried out according to the analysis of the actual situation of C company,and risk sources and influencing factors are found out.Secondly,the risk sources and influencing factors are used to determine risk early warning indicators,and the weight method of CRITIC is introduced to calculate the weight of indicators.Finally,the FCM clustering algorithm is used to evaluate the risk of Company C,and four risk levels are evaluated.The evaluated data are trained by the random forest algorithm and used to predict the risk points,so as to obtain the final prediction accuracy.Since there are four risk levels in the random forest predictable risk assessment,this paper proposes corresponding prevention and control strategies according to the four types of risks that can be predicted and the risk sources identified by C Company.The results show that:(1)After using FCM clustering algorithm to evaluate the risk of data points of Company C,there are significant differences among different categories of data points,indicating that FCM clustering algorithm can effectively evaluate the data points of Company C;(2)Risk prediction based on random forest algorithm has a high accuracy and can effectively predict risk points;(3)The weight judgment of indicator by CRITIC weight method has not affected the prediction accuracy,which shows that it is effective to judge indicator weight by similarity and volatility.At the same time,FCM clustering algorithm and random forest algorithm are introduced into the risk early warning system to identify,evaluate,forecast,prevent and control the company’s operating risks in conjunction with the company’s historical data and traditional risk management system.The results show that this kind of method is effective and can provide guidance for the relevant companies to establish the risk early warning system,which has important practical significance.