基于Fisher判别法的P2P网络借贷平台信誉等级评价模型

The P2P Lending Platform Credit Rating Evaluation based on the Fisher Dicscriminant Model

P2P网络借贷是一种新兴的互联网金融,平台的信誉是影响投资者选择平台的主要影响因素之一。选取成交积分、人气积分、营收积分、分散积分、杠杆积分、透明度、品牌、流动性、收益积分9个影响因素作为判别平台等级预测的评价指标,同时对这9个评价指标做主成分分析,提取出3个主要成分,用Fisher判别法进行训练预测,建立了基于主成分分析的网络借贷平台信誉等级预测的Fisher判别模型。通过对"网贷之家"公布的37组平台数据作为训练样本数据集进行模型的训练,12组数据作为该预测模型的测试数据,进行网络借贷平台信誉等级的预测,同时通过其他预测模型预测结果的对比,验证了Fisher判别法在网络借贷平台信誉等级的预测中具有较低的误判率,其误判率仅为1/12。

P2P lending is a kind of emerging Internet finance, the reliability of the platform is the mainfactor affecting the investors chooses the platform. This article selects volume integral, sentiment, revenueintegral, scattered integral, integral liquidity, transparency, brand, leverage, earnings integral ninediscriminant platform level forecast factors as evaluation index, evaluation indexes of the nine familycomponent analysis, to extract the three main ingredients, trained prediction with the Fisher discriminantmethod, based on principal component analysis of the Fisher discriminant model of network platformfor lending credit rating. Released by means of ＂net house＂ of the 37 group platform data as thetraining sample data sets model of training, 12 groups of data as test data, the prediction model for predictionof network platform for lending credit rating by comparison with other prediction model to predictthe result at the same time, fisher discriminant method is verified in the network platform for lendingcredit rating prediction with low misjudgment rate, the miscarriage rate is only 1/12.