Credit Evaluation Model Based on Time Series Fuzzy Clustering and Rule Extraction
Zhizhong Mao · Journal of Northeastern University · 2010
A fuzzy classification system was presented by combining the multidimensional time series fuzzy clustering with rule extraction to evaluate the credit.This method was applied to the dimensionality reduction of multidimensional time series via projection pursuit technique for fuzzy clustering.According to the results of the classification and the extraction of fuzzy rules through best value projection,three types of fuzzy membership function were generated using trapezoid distribution method.Then the sample's credit level was determined according to the fuzzy approximability in computation.Exemplification results showed that this method is good at evaluation effectiveness and practicality.