A Fuzzy Random Forest And Its Application In Enterprise Financial Risk Early Warning

With the continuous and rapid development of China's economy,the operating environment of'listed companies has become more and more complex,coupled with the increasing international competitive pressure of enterprises,the financial risk of listed companies need more attention.If we do not pay attention to the financial risk status of the enterprise,it will lead to the continuous accumulation of financial risks and eventually cause financial crisis,thus being marked by ST.Financial crisis is gradually manifested in finance from the perspective of its generation process,so we can use the financial index system to build a model to make early warning of enterprise financial crisis.In terms of the research status of financial risk early warning,in order to improve the accuracy of financial risk early warning,many researchers have carried out long-term research and exploration,and formed the early warning method represented by univariate model and multivariate model.At present,there are many financial risk warning models mentioned,but most of them have the following problems:generally,insufficient training leads to insufficient learning or over-fitting;There are many parameters to be adjusted,and the training cost rises sharply with the increase of training samples.Only fixed-point prediction can be achieved,and the predicted results cannot be further divided.Aiming at the above problems,a fuzzy random forest model combining fuzzy mathematical theory and random fotest is proposed.Random forests(Random Forest,RF)is represented in the field of data mining algorithm,can be dug up a lot of information from the limited data,RF algorithm for Bootsrap heavy sampling method to obtain the training sample,its basic idea is to construct the decision tree model,has high forecast precision and generalization error control,fast convergence and few parameters adjustment,can effectively avoid over fitting phenomenon,especially for high-dimensional data operation.By combining fuzzy mathematics to transform the binary classification data into multi-classification data,and then input it into the random forest model for training and prediction,a fuzzy random forest model that can do multi-classification prediction can be obtained,so as to further dig deeper information in the sample.This paper studies the theoretical knowledge of stochastic forest algorithm,improves the stochastic forest algorithm with fuzzy mathematics,and then applies the stochastic forest algorithm with fuzzy mathematics in the field of enterprise financial risk warning.The main work is as follows:1)from the Internet to climb took the financial data of a-share listed companies from 2013 to 2017,and calculated the ratio of financial indicators,using the quarterly data,relative to other scholars in the field of enterprise financial risk early warning using the annual report data,this paper expanded the amount of data to its 4 times,the increase of the amount of data is also made in this paper,the model of building more accurate.2)this paper applies the random forest algorithm to the early-warning field of enterprise financial risk for the first time,and proposes the fuzzy random forest by introducing fuzzy mathematical theory and borrowing the ideas of Bagging method and Breiman random forest method.This method overcomes the defect that traditional financial forecasting model can only get dichotomous results and can get more detailed financial risk classification.