Partition Ratepaying Credit Grade Based On Data Mining

This paper focuses on how to partition ratepaying credit grade based on data mining and how to develop system prototype about partitioning ratepaying credit grade. The paper pinpoint that it is necessary to analyze ratepaying credit, expound foreign and home research status and significance of it and knowledge about data mining. Subsequently, credit and evaluating model are stressed. Aiming at partition ratepaying credit grade, this paper sums up that ANN model is very adapt to evaluate credit grade.The paper discuss outlier detect and cluster technology in data mining. We propose a new mining approach--outlier-based clustering analysis. This algorithm can not only mine outlier data, but also improve cluster's precision. At the same time, it plays an important role in data pre-processed and train ANN. The BP algorithm for feed—back neural network is researched and applied abroadly in the practice, however, it has a set of shortage. Slow convergence and easy to get into local least point are the main problem. We put forward two methods to optimize BP algorithm. This paper sets forth SPDS algorithm based on non-grads to get over BP's lack. The experiment shows that SPDS have better performance compared with BP algorithm, for example spate and tolerance-mistake ability. Therefore SPDS is suitful to practical application.At the last we apply neural network based on SPDS to partition ratepaying credit grade. We extract processed data as training set from database of Taiping tariff and produce classsify based on neural network. These methods have enough advantage compared traditional means. Thinking the task of partition ratepaying credit grade in Taipingtariff, we designed a system prototype based on data mining to partition rateplaying credit grade, which is valuable for next step.