Construction And Application Of Goodwill Impairment Crisis Early Warning Model Based On CART Algorithm

Since the performance explosion caused by the impairment of the goodwill of listed companies in 2018,the market is still releasing the huge amount of goodwill accumulated by the M&A frenzy before 2016.The risk of impairment of goodwill and goodwill becomes the "sword of Damocles" hanging high above the heads of listed companies.It is not uncommon for listed companies to face the risk of delisting due to the loss of net profit due to the impairment of huge amounts of goodwill,The healthy development of China’s capital market has been significantly threatened by the impairment of goodwill,and how to alleviate the systemic risks and investment risks that may be brought about by the huge impairment of goodwill has become the focus of attention of regulators and investors.In order to predict the future goodwill impairment crisis of the enterprise in advance and avoid the performance "thunderstorm" caused by the impairment of goodwill,we need to establish an effective goodwill impairment crisis early warning system.It has very important practical value and practical significance for listed companies and other stakeholders.This paper takes the GEM,which is active in M&A market transactions and the "hardest hit area" of goodwill impairment,as an example to study the problem of goodwill impairment risk prevention.The net profit of the listed company due to the impairment of goodwill is negative as the criterion for judging the risk of impairment,On the basis of theoretical analysis,this paper selects the sample data of the GEM from 2017 to 2020,this paper uses the CART algorithm to construct an early warning model of goodwill impairment crisis,The importance of each feature variable is derived,as well as a visual graph of the decision tree.According to the data of information transmission,software and information technology service industry,102 companies in the entire industry in 2020 were predicted,and compared with the real results,99 companies were accurately predicted,and 3 companies were mistakenly judged to have a goodwill impairment crisis,and the accuracy of the forecast results was 97.06%.Then,the reasons for the error were analyzed,and among the 4companies that did have a goodwill impairment crisis,Gao Weida Company with the highest amount of goodwill impairment was selected to conduct a detailed analysis from three perspectives: the formation of goodwill impairment,the visual analysis of decision trees,and the analysis of feature importance indicators.Finally,the recommendations were made from the three levels of internal company,government supervision and investors.The research contribution of this paper is to generate an evaluation table of feature materiality indicators,which quantifies the importance of feature indicators to the goodwill impairment crisis affecting enterprises;Generate a visual diagram of the decision tree to display the specific classification rules of the model;And through the construction of models,help enterprises predict whether there will be a goodwill impairment crisis in one year,so that enterprises can actively take action in the following year to promote risk prevention.