Selection Of Variables And Indicators In Financial Distress Prediction Model-Svm Method Based On Sparse Principal Component Analysis

Abstract

How to screen out the early warning indicators from a large number of alternative financial indicators is an important step in the prediction of financial distress. In order to design the financial distress prediction model more effectively, this paper proposes a novel method that combines the sparse algorithm and support vector machine. Firstly, according to the financial management theory, financial indicators are divided into several groups, and then variable screening was conducted for each group of financial indicators by sparse principal component analysis. Finally, after variable screening, the data set is input to the SVM for classification and prediction (classifier prediction). The empirical results show that this method can more effectively identify companies in financial distress, improve the prediction results of the model, and reduce the risk of investors facing financial distress.