Projection Pursuit Cluster Analysis And Its Application On Financial Performance Appraisal Of Listed Companies

At present, the listed companies are becoming the major-micro-part of China's economy which makes the integrative appraisal and analysis of its profit-making state become popular projects.In the research of the financial performance appraisal of the listed companies, there are multifarious integrative appraisal methods, such as subjective weighting methods, factor analysis, entropy weight methods and so on. However, these appraisal methods itself have many problems, such as overmuch disturbance of factitious factors, contradiction between data length and estimate precision which is called "The High-dimension Problem" in statistics and so on. Those induce that appraisal results are not objective and robust, and cannot truly reflect the status of the financial performance of the listed companies. It brings difficulties to the listed companies' supervision and appraisal of the listed companies' quality from investors. Therefore, it is very significant to develop theory of enterprise performance appraisal, to supervise the listed companies and to make decisions for investor that exploring a more objective and more robust appraisal methods than other existing appraisal methods from the financial data itself of the listed companies.On the basis of above cognition, based on the all-around review of the existing integrative appraisal method of the listed companies, this article tries to apply the Projection Pursuit Cluster Analysis (PPCA) to the research of the financial performance appraisal of the listed companies which is a multivariate statistical method that produced from natural science field and resolve the high-dimension problem. We choose the data of the eighteen listed companies in Shanghai stock markets and Shenzhen stock markets, apply the Real-coded Accelerating Genetic Algorithm (RAGA) to resolve the key problem of optimal projective direction in PPCA, and empirical study the status of the financial performance of eighteen listed companies. To see about validity and practicability of the financial performance appraisal method based on PPCA, this article applies twelve exiting integrative appraisal methods to appraise the status of the financial performance of eighteen listed companies. Then further applying Sequence-number Summation Theory and Mode Theory, to check up the appraisal precision difference between appraisal results based on PPCA method and exiting methods.The research results indicate that applying Sequence-number Summation Theory or Mode Theory for testing method, the financial performance appraisal method based on the Projection Pursuit Cluster Analysis performs better appraisal precision than exiting appraisal method as a whole. At the same time, this research discovers appraisal results based on PPCA after the Normal Standardization which can obtain more outstanding appraisal precision by these two testing methods. Thus, applying Normal Standardization Method to deal with data is more logical and advisable choice. Because of considering multifarious integrative appraisal methods in this research, and applying two different testing methods with Sequence -number Summation Theory or Mode Theory, the research results have comparative robustness and practicability.The research provides a fire-new angle of view and a technical means for the financial performance appraisal of the listed companies, and provides an effective method choice for investor and supervisor to appraise the financial performance of the listed companies.