Research On Financial Fraud Recognition Of Listed Companies Based On Machine Learning

In recent years,with the development of the capital market,financial fraud of listed companies in China has emerged one after another,and the fraud methods have shown the characteristics of concealment and complexity.How to effectively identify and prevent financial fraud and stability of listed companies is of great significance to the healthy development of the capital market.The development of big data technology provides an effective means for processing massive data and extracting valuable information-machine learning methods,with the help of machine learning methods can gain insight into the patterns and correlations of internal factors in the data information of listed companies,quickly discover the laws of related problems,timely reveal the financial fraud of listed companies,and achieve more efficient,objective and accurate identification of financial fraud of listed companies.This paper first summarizes the relevant research on the means,identification characteristics and identification methods of financial fraud,and lays a foundation for the preliminary selection of characteristic indicators for identifying financial fraud of listed companies based on relevant fraud theories.Then,on the basis of the data mining process,a process framework based on machine learning financial fraud identification method is proposed,including three steps: sample and feature selection,data preparation,and model construction and analysis.At the same time,this paper introduces the drawbacks of traditional financial fraud identification methods and the characteristics of machine learning to identify financial fraud.Based on the process framework for financial fraud identification method based on machine learning,this paper takes the manufacturing industry as an example,selects characteristic indicators from nine levels,selects listed companies with financial fraud in 2010～2021 as fraud samples,and integrates,cleanses,balances and other processing of the obtained data,and obtains the data that can finally be used for input algorithms.Based on XGboost algorithm and light GBM algorithm,the financial fraud identification model is constructed.The model is analyzed and evaluated,and the contribution of each feature to the financial fraud recognition is determined according to the importance of the features,and the feature indicators containing the largest amount of information are screened out to establish the final model.Eight listed manufacturing companies with financial fraud were selected from the typical illegal cases released by the CSRC,and the constructed models were used to predict the financial fraud of these companies to verify the accuracy and reliability of the models.The results showed that the model constructed in this paper had a good identification effect.Taking the manufacturing industry as an example,a financial fraud identification model of listed companies is constructed based on machine learning methods,intelligently identifies fraud behavior of listed companies,realizes early warning and prevention,and provides a more effective means for establishing an effective early warning mechanism.It enriches the theoretical research on intelligent identification of financial fraud,has guiding significance for the construction of financial fraud identification model of listed companies in the era of big data,and provides ideas for the establishment of intelligent fraud identification system.