Alternative Scoring Factors using Non-Financial Data for Credit Decisions in Agricultural Microfinance
Alternative Scoring Factors using Non-Financial Data for Credit Decisions in Agricultural Microfinance
复制标题
使用非财务数据进行农业小额信贷信贷决策的替代评分因素
DOI：
10.1109/syseng.2018.8544442
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
N. Kohtake
中科院分区：
文献类型：
--
作者：
Naomi Simumba;Suguru Okami;A. Kodaka;N. Kohtake
金融排斥对穷人和无银行账户的人有重大的社会经济影响。财政上被排除在外的小农面临着获得信贷设施为其农业活动提供资金的挑战，因为他们缺乏为信用风险评估创建信用评分所需的财务历史数据。对于信用评分模型的开发，已经提出了非金融数据源，如移动应用程序。然而，对于使用非金融数据的信用决策系统，必须开发独立于财务历史信息的特定于上下文的替代评分因素。这项研究提出了一种根据利益相关者的要求制定替代评分因素的方法。使用通过调查和移动应用程序从柬埔寨农村农民那里收集的数据，给出了实施建议方法的实例。根据利益相关者的要求和收集到的数据，制定了可供选择的评分因素。在此基础上对多元Logistic回归模型和支持向量机模型进行了训练和测试，以评价所选因素。对模型按面积下接收器的工作特性、曲线值和精度进行了比较。为了确定在这种情况下最合适的模型，还进行了其他考虑。这种基于利益相关者需求的方法可以用来设计信用决策系统，使用财务上被排除在外的人的非财务数据，并促进更大的财务包容性。
Financial exclusion has a major socio-economic impact on the poor and unbanked. Financially excluded smallholder farmers face challenges accessing credit facilities to fund their farming activities because they lack the financial history data required to create credit scores for credit risk evaluations. Non-financial data sources such as mobile applications have been proposed for the development of credit scoring models. However, context-specific alternative scoring factors which are independent of financial history information, must be developed for credit decision systems that use nonfinancial data. This research proposes an approach to developing alternative scoring factors based on stakeholder's requirements. An example of implementation of the proposed method is given using data collected from farmers in rural Cambodia through surveys and a mobile application. Alternative scoring factors are developed based on stakeholder's requirements and collected data. Multiple logistic regression and support vector machine models are trained and tested on this data to evaluate the selected factors. Models are compared by area under the receiver operating characteristics curve values and accuracy. Additional considerations are made to determine the most suitable model in this context. This stakeholder requirements-based approach can be used to design credit decision systems using nonfinancial data for financially excluded persons and facilitate greater financial inclusion.