Financial crisis early-warning based on support vector machine

Abstract

Analyzing the principle of typical financial crisis early-warning model, this study summarizes the limitations of them and their requirement of variance. An empirical research is carried out on how to sample the Chinese listed companies of A-stock market in Shanghai and Shenzhen, and how to determine the core parameters of support vector machine (SVM) as well. This research also studies the predicting accuracy in 1-3 years and the performance on condition that some data are missing. At last the contrastivAnalyzing the principle of typical financial crisis early-warning model, this study summarizes the limitations of them and their requirement of variance. An empirical research is carried out on how to sample the Chinese listed companies of A-stock market in Shanghai and Shenzhen, and how to determine the core parameters of support vector machine (SVM) as well. This research also studies the predicting accuracy in 1-3 years and the performance on condition that some data are missing. At last the contrastive analysis is made between SVM model and the Logistic model. Our experimentation results demonstrate that SVM outperforms the logistic model and SVM also has a sound accuracy under the data missing.e analysis is made between SVM model and the Logistic model. Our experimentation results demonstrate that SVM outperforms the logistic model and SVM also has a sound accuracy under the data missing.