MACHINE-LEARNING CREDIT-SCORING FOR UNBANKED MICRO-RETAILERS: A LASSO-LOGIT ENSEMBLE APPROACH

Abstract

In the Third World, basically only micro-tables are shut out from loans because they don't have financial identities. The research focuses on developing a trustworthy credit-scoring model for micro business who remain unbanked. This study expands on previous research that uses small loans, but doesn't know the details on the consumer's risk level and repayment. By combining payment history along with another alternative method, micro-retailers can be accessed whether they are reliable on payments or not. No decision was made because conditions weren't in place for there to be a test to be conducted, possibly for the second amount or on June 24 (yield). The report introduces numerous ways of securing alternative credit that let's everyone join the community including those without access to certain groups. Machine learning could have a strong influence on the economy's financial system, however not equally affecting those in the business class and middle class.

Keywords: Machine Learning, Credit Scoring, Micro-retailers, LASSO, Logistic Regression, Financial Inclusion, Unbanked.