Early warning model of the financial crisis; An empirical study based on partnering of biotech firms in china

Abstract

The overfitting phenomenon in the financial crisis early warning model leads to low accuracy. A financial crisis early warning model is established based on the partnering of Chinese Biotech firms. Starting from the four sources of the crisis, use indicators to quantitatively reflect the main problems and establish a financial crisis early warning index system. According to the characteristics of financial data, a rough set is used to discretize the data to make the data characteristics clearer. The financial crisis is classified based on a maximum entropy clustering algorithm to overcome the overfitting problem. Calculate the index’s weight, obtain the comprehensive score, judge the crisis category, and complete the establishment of the financial crisis early warning model. The experimental results show that the average accuracy of the financial crisis early warning model based on MEC is 0.9290, which is 0.1442 and 0.1697 higher than that based on the SVM and RBF neural network. Therefore, the judgment accuracy of this model is much higher than that of SVM-based and RBF-based early warning models, and the results are mostly in line with the actual situation.