Quality Assessment For The Results Of Funding Rating Based On Data Mining

Credit rating is a management activities doing by independent rating agencies or professional sector.By taking scientific evaluation methods and evaluation procedures,they intent to use the symbols to conduct a comprehensive evaluation of risk factors.Fund rating is an important part of credit rating.It is done by the rating agencies based on established standards.With the development of funds and other securities market,third-party rating agencies increase.More and more people put their eyes on the results of rating agencies.The quality of rating results is the ability to promote the development of the fund industry.Home and abroad have various methods for quality inspection of the ratings,like default rates test,test the stability of the migration matrix,the Gini method,the rating distribution of test results,z scoring models and ratings correspondence between the inspection.This paper presents a method to evaluate a fund rating result based on data mining.First,analyze the impact of fund performance using factor analysis method to extract the effective information;and then use cluster analysis algorithms to classify the funds;finally using rating accuracy,false positive error rate and average ratings and other indicators to show the quality of the ratings.In this paper,we use the data of April 2011 stock fund ratings to study,use C++programming language to implement PAM algorithm.By using rating effectiveness and stability test,the result is Galaxy Securities stays strong rating,rating better,followed by Morningstar,Ji An Jinxin last.