Intelligent Financial Risk Identification System Supported by Enterprise Financial Data Mining in Big Data Environment

Abstract

In this paper, 27 corporate financial risk indicators are first selected in the big data environment, and 8 non-significant indicators are screened out from the 27 indicators by using K-W test and T-test; and then further condensed into 6 factors by combining with the factor analysis method and constructing the financial risk early warning indicator system. Then, PSO-BP neural network based on particle swarm optimization algorithm is constructed, and the identification effect of this network in enterprise financial risk is simulated and tested. The results show that: after screening, this paper finally selected 19 indicators as the early warning indicators of the model; after factor analysis identification of the 19 indicators, six main factors are obtained, such as dominated by the net equity interest rate and dominated by the balance sheet ratio, whose response rate to the financial characteristics of the sample enterprise is more than 75%. The training number of PSO-BP neural network reaches 21 times, then it can reach the prediction accuracy set by this paper, and then the overall fitting effect of the model is tested in simulation. Accuracy, at this time, the overall fit R value of the model is as high as 0.9871, the network prediction error range is between -0.39-0.31, the accuracy of the expected value and the predicted value of the financial risk identification is more than 95%; the accuracy of the identification of the enterprise's financial data in the big data environment reaches 98.15%, which efficiently completes the intelligent identification of the enterprise's financial risk.