<![CDATA[Financial institutions tend to prefer to have borrowers with a clear source of credit history. Unlike P2P Lending which does not have complete data, P2P Lending uses alternative or substitute data to create a credit scoring model. Social media as one of the alternative data options cannot directly translate content related credit problems. Therefore, it is necessary to have an approach and understanding of social media data. By using data user demographic attributes or social media demographic features, this study aims to create a credit scoring model using substitute data in the form of social media data for P2P Lending. This study uses a data mining process that utilizes a decision tree algorithm and random forest. The results showed that the credit scoring model with the random forest algorithm produced the highest accuracy of 81.25% and was chosen to be the best model. From these results, P2P Lending can have the opportunity to open a new segment of consumers who do not have a complete financial or credit history.]]>
Copyrights © 2020
