GM(1,1)模型的改进及其在银行贷款预测中的应用

Improving on GM(1,1) and Its Application in the Forecast of Bank Loan

利用GM(1,1)预测模型,分析中国近年来信贷变化,发现中国信贷规模发展趋势及其信贷结构具有显著的新特征;结合模型的适用条件和预测结果的检验要求,提出了四条思路解决数据缺损导致信息挖掘失真以及如何处理数据中所包含的大量"噪音"等问题,并认为对那些波动较大的数据应采取连续平滑的方法改造原始数据;对于那些有奇异值的原始数据可采用插值法替代它,试验和案例应用印证了改进的模型可以提高预测精度.

GM （1,1） is a kind of grey prediction model that extracts valuable information through producing and exploring ＇some＇ known information. This paper utilizes this model to predict and analyze the changes of credit in China in recent years. It finds out the development trend of credit scale and the significant new characteristics of credit structure. In addition, it also notices that there are some limits exist in GM （1,1） model in utilizing the primitive data. For this reason, combining the terms that is suitable for the model and the test requirements of prediction result, the paper proposes four schemes to give solutions to problems such as the data deficiencies that results in information excavating distortion and the ＇noises＇ included in the data processing. Moreover, it believes that for data which have considerable fluctuation, the continuous smoothing method should be adopted to transform primitive data; while regarding those primitive data which have singular values, if the singular values lie in the data, then internal inserting method can be adopted to substitute them ; if the singular values are the latest data, then the external inserting method can be used to substitute them, and therefore set up new primitive data sequence and improve conditions that are suitable for the model, it is confirmed by test and application of case study that the precision of prediction can be improved by such improvements.