Research On The Measurement Method Of Generalized Metric Matrix And The Application Of Credit Rating

The main research direction of credit rating is building the credit rating model.Normally, the credit rating model isn’t always right, there are also some problems.Therefore, building a more scientific and rational model is very meaningful. The differences between the credit rating have always been there.The distance is often used as a measure of the different cluster between two kinds of samples, and also as the criteria for the classification of the sample. In the study, the unit matrix based on Euclidean distance and the covariance matrix based on mahalanobis distance is the most widely application scope of two kinds of measurement scales, however there are a lot of inapplicability of these two kinds of measurement scales. If we can find a more reasonable measurement scale in order to describe the distance difference or estimate the generalized distance metric matrix. In the next study, we can directly use the metric matrix to measure the distance between one cluster of questions or sample meaningfully. At the same time,the conclusion we make by the measurement matrix eigenvalue decomposition can be objectively expressed the signification of each index, avoids the subjective empowerment as many research methods used in the correlative questions. In the study of generalized measurement matrix measure,it has been divided into the following two main research directions: directly use the differences between the sample concentration of each sample to measure metric to build the rating model; And use the extreme value theory, with the classification results of each sample we know and the sample concentration index variables for every average optimal distance as the judgement standard and calculate measure matrix and to build the rating model. This paper divided into two types of generalized metric measurement formula, mainly using the generalized distance measure expansion, based on two research directions of measurement matrix calculated and the complete mathematical derivation has been given. The method of mathematical methods is mainly matrix algebra, specific to the two research directions,using the matrix least square method and the Lagrange multiplier method.The generalized metric matrix is derived in this paper, a generalized metric evaluation formula has been constructed and empirical tested.We make the conclusion that the new model with high precision is suitable for the application of the research on credit rating.