Credit Evaluation Based On The Multidimensional Time Series Modeling And Its Applications

With the development of research methods for the credit evaluation, more historical data (time-series data) that reflect company characteristics are used in credit evaluation. However, time-series data often exhibit the characteristics of nonlinearity, chance and time variability, which increases complexity to establish the credit evaluation models. In particular, the results of the credit evaluation are difficult to guarantee accuracy. How to effectively analyze, use time-series data and provide accurate credit assessment information for decision-makers is one of the central issues recently.Chinese enterprises have relatively short development history, and most of the evaluation data are collected from the corporate financial statements (the quarterly or annual). Due to less time-series sampling data for credit evaluation and the characteristics of time-series data make modeling more difficult. To solve above issues in credit evaluation, some algorithms and solutions of credit evaluation are proposed based on research results at home and abroad and then the effectiveness of the proposed method is verified by computer simulation. The main ideas in this dissertation are listed as follows:(1) To solve the problem that isolated point in time-series data easily lead to distortion of evaluation results. The multidimensional time series credit evaluation model based on the gray fuzzy is proposed. This method can evaluate objective assessment credit rating within the sampling period and obtain stable credit trends associated with objective evaluation results, which improves control and avoidance of credit risk.(2) The evaluation index attributes have a great significance on credit evaluation. Especially when the index attributes are less and highly targeted, they have larger impact on the evaluation results. In response to this situation, a credit evaluation model based on fuzzy MADM multivariate time series is proposes. According to the characteristics of the multi-dimensional time-series data, the method uses deviation maximization model and quadratic programming model to derive weights of index attributes as decision parameters used in the modelling of credit evaluation.(3) Fuzzy cluster analysis is mainly used to solve the problem that credit category of sample is unknown. It makes subsequent external sample evaluation results are unsatisfactory, which leads to certain limitations in the practical application. To solve this problem, a credit evaluation modeling method of multidimensional time series based on fuzzy clustering and rules extraction is proposed. By combining fuzzy clustering analysis and rule extraction, evaluation rules are extracted to establish the credit assessment and classification function, which extends the application scope of model in credit evaluation.(4) The introduction of time-series data to credit evaluation, the predictability of the time-series provides a theoretical support for the prediction of credit risk. The credit evaluation prediction model based on multi-dimensional time-series is proposed. From the sampling period point of view, the model evaluates and predicts credit rating.