基于数据可视化Rattle的个人信用风险评价建模

Modeling of Personal Credit Risk Evaluation Based on Data Visualization Rattle Tool

本文从个人信贷的定义出发,先总结了国内外个人信贷评估的经验理论,再根据模型评价和综合评判两个方面进行模式创新,通过实证研究,挖掘出了决定我国个人信用状况的重要变量,通过数据可视化Rattle建立起个人信用评价模型。利用SVM、决策树、Neural Net等算法进行有效性的比较,又利用Logistic回归这一经典计量模型扩展了客户进行信用评分。除此之外,借鉴VaR概念提出Profit-at-Risk模型,测算出使用模型存在的风险,并且在收益和风险中取得平衡,弥补了数据挖掘模型只关注准确度的不足。

From the definition of personal credit,first summarizes the experience of personal credit evaluation theory at home and abroad,then the mode of innovation according to the two aspects of model evaluation and comprehensive evaluation,through empirical research,dig out the important variables in determining our personal credit situation,through the visualization of data Rattle to establish personal credit evaluation model.Using SVM,decision tree,Neural Net and other algorithms to compare the effectiveness,and the use of Logistic regression classical econometric model to expand the customer credit scoring.In addition,the Profit-at-Risk model is put forward by using the concept of VaR,and the risk of using the model is calculated,and the balance between revenue and risk is made up,which makes up for the lack of accuracy in data mining model.