Title: Profit scoring in peer to peer lending
Authors: Trevor Fitzpatrick - University of Southampton (United Kingdom) [presenting]
Abstract: The purpose is profit scoring of Peer to Peer loans. To accomplish this, a comparison of ensemble algorithms such as Regularised Random Forest and Regularised Regression is carried out. Transformations of the response variable are also investigated. The results suggest that rank-based transformations and ensemble-based regressions produce the most promising results.