Study On Small Business Credit Scoring And Its Application Based On Logistics-ANN Model

Since the Mid-1990s exploration on small business credits has been a hot research area on a wide-spread basis. The related scholars have done comprehensive researches on small business credit scoring from different angles and ranges, which has not only formed the theoretical results but also promoted the development of local economy. In foreign countries, small business has its own methods and indicator systems for credit scoring which suit to their own characteristics. But in China study on small business credit scoring is still at initial stage, and the majority tend to mix small businesses and medium-sized together, which do not match characteristics of small businesses in China. Furthermore, there have been very few literatures specifically for small businesses in domestic study, and it has not yet formed a complete evaluation system as well as method for small business credit scoring.In view of the current research of small business credit scoring and based on domestic and international research literatures, the study began with analysis of the characteristics of small businesses and the existing research methods together with its application results to analyze the credit data and indicators, drawing the conclusion that the existing research methods have inadequacies and defects in small business credit scoring. For this reason, this paper constructed a model consisting of logistics analysis and artificial neural network analysis namely Logistics-ANN. Firstly introduce the logistics model to select the indicators which affect the small business credit according to characteristics of small businesses. Secondly use the selected significant variables from logistics model act as the input variables for neural network model to conduct the credit scoring, followed by the in-depth study of application of this Logistics-ANN model in small business credit scoring. Finally, compare the results of the applications with the existing methods and draw the conclusion that the Logistics-ANN model has advantages of objectiveness, high accuracy and simple etc., and then come up with conclusions and recommendations on the further study of small business credit scoring.