The Application of Internet Big Data and Support Vector Machine in Risk Warning
Jiang Li,Yaochen Tan,Ao Zhang
DOI: https://doi.org/10.1088/1742-6596/1952/4/042026
2021-06-01
Journal of Physics: Conference Series
Abstract:Abstract With the strengthening of macro-control of the real estate industry and the intensification of market competition, it is of great significance for the steady development of the industry to establish an accurate and effective early-warning mechanism for enterprise financing risk. Taking the A-share real estate listed companies in China in 2019 as target, this paper collects financial information of the relevant companies from 2010 to 2019, supplemented the risk sample data from 2005 to 2010, and reduced classification imbalance by removing discrete points and using SMOTE. When the financing risk evaluation system of listed real estate companies is constructed with capital as the entry point, the stochastic forest algorithm is used to finally select five important characteristic dimensions, namely, current ratio, equity financing ratio, operating income, current liability ratio and receivable turnover ratio. This paper establishes SVM early warning model, PSO-GA-BP optimization model and KNN early warning model to predict financing risks of real estate companies. It is concluded that, by comparing their applicability and advantages and disadvantages, SVM after feature screening and sample processing has better performance in the real estate financing risk early warning, which can provide some references for enterprise decision makers, investors and regulatory authorities.