Prediction Of The Risk Level Of Agricultural Loans For Agricultural Bank Of China Based On Support Vector Machine

China is a large agricultural country, "three rural issues" has been the relationship between China's economic progress and social stability. But China's rural economic level is low, the cultural quality of farmers seriously inadequate, resulting in China's agricultural development scale, modernization level can not keep up with the need of development of era. "Three rural" loans is a help to the development of "three agriculture" Huinong policy of the government and financial institutions to jointly launch, can promote the scale, the development of the industrialization of agriculture to a certain extent. However, because our country fanner loan risk resisting ability seriously insufficient, leading to frequent adverse credit phenomenon, so that financial institutions are too cautious in the lending process, leading to the "three rural" loans unreasonable distribution and lending is not timely, which in turn restricts fanner's repayment ability, so that the "three rural" loan limit deadlock.In order to make the farmers in the "three rural" in the process of loan and bank loans reached a win-win situation, to promote the "three rural" sustainable development loans, through statistical analysis of Changsha City, a China agricultural "three rural" loans first data, using correlation analysis and principal component variables selection on the prediction data analysis method, and then based on machine learning method support vector machine has established a "prediction model of three rural" loans two customer classification model and the "three rural" five level loan risk rating.Validation of a Chinese from Changsha City Agricultural Bank of a random sample of 500 "three rural" loans customer information to the model established in this paper, found that the support vector machine based on the "three rural" loans two customer classification model classification accuracy is slightly higher than that of BP neural network in classification accuracy, remember "support vector classification accuracy prediction model of agriculture" five level loan risk level of precision is higher than that based on BP neural network, prediction accuracy of this model shows higher, better stability, shorter training time and support vector machine, high efficiency. This paper studies the sustainable development is of great significance for the "three rural" loans.