Associating grey relation and cluster analysis to perform financial characteristic study on enterprises in China

Abstract

In recent years, global economic environmental are facing a fluctuated and unstable period. Due to the influence of subprime mortage crisis in USA and debt crisis in Europe, bankdruptcy of many large enterprises occurs one after another. In many countries, credit bankruptcy and violation of debt contract also occurs frequently, the violation of settlement in stock market can even be seen occasionally. This indicates that for many companies, not only risk is not well managed, but also the risk they are facing with is not well understood. To avoid possible business operation risk, the managers of a company really need to inspect the financial situation and characteristic of a company in advance. In this study, first, profitability force and growth force of financial five forces are used to the financial ratio data collected from the enterprises of China. Meanwhile, stability force and activity force is used to perform Grey Relational Analysis and the analysis results are sorted according to grey relational grade. Moreover, PSO (Particle Swarm Optimization) Clustering algorithm is adopted to categorize companies into two groups. Financial characteristics of these two groups are studied and the results are useful for the bank managers. In this study, three data mining techniques: Genetic Programming (GP), Back-Propagation Neural Network and Logistic Regression are used to build the financial characteristic detection model of an enterprise. From the analysis result, it can be seen that PSO cluster analysis divide the enterprise into two groups, one group with good stability and activity force, another group with inferior stability and activity force. However, among these enterprise financial detection models, the classification forecast capability of GP model is the best.