BP神经网络在农户信用评级领域的应用研究——以甘南藏族自治州为例  

Application Research of BP Neural Network in the Field of Farmers’ Credit Rating——Taking Gannan Tibetan Autonomous Prefecture as an Example

作　　者：课题组 梁荣[2] 严小军[2] 布贵臣[2] 冯华 LIANG Rong;YAN Xiaojun;BU Guichen;FENG Hua(Gannan Municipal Sub-branch PBC,Gannan Gansu 747000)

机构地区：[1]不详 [2]中国人民银行甘南州中心支行

出　　处：《西部金融》2019年第11期88-92,共5页West China Finance

摘　　要：近年来,随着银行信贷经营面临的风险更趋复杂化和信贷精细化管理要求的逐步提高,如何通过客户和债项二个维度,综合判断农户贷款偿还的可能性,是目前急需研究的课题。本文以甘南州8家农村信用社的300户农户数据为样本,从农户家庭特征、收入支出水平、资产负债情况和社会评价四个方面选取18个影响农户信用的指标与贷款五级分类构建BP神经网络模型,通过对隐藏节点个数的变化选择以及神经网络模型相关参数的设置调试,最终运用十四层BP神经网络对样本农户信用状况进行综合评价和验证,推动涉农金融机构农户信贷评级应用,降低农村信贷风险。In recent years,with the increasing complexity of bank credit operations and the gradual improvement of credit fine management requirements,how to comprehensively judge the loan repayment possibility through the two dimensions of customers and debts is an urgent issue.This paper takes the data of 300 households in 8 rural credit cooperatives in Gannan as a sample,and selects18 indicators that affect the credit of farmers and the five-category of loans from the four aspects of household characteristics,income and expenditure,asset and liability,and social evaluation to bulid the neural network model,through the selection of the number of hidden nodes and the setting and debugging of the relevant parameters of the neural network model,this paper finally uses the fourteen-layer BP neural network to comprehensively evaluate and verify the credit status of the sample farmers,and promote the credit rating of the farmers in the agriculture-related financial institutions.Application to reduce rural credit risk.

关 键 词：农户信用评价 BP神经网络 贷款五级分类 

分 类 号：F832.4[经济管理—金融学]