The objective of the dissertation is to propose an explainable credit risk assessment to examine the explanatory variables in P2P lending platform and make a comparison between the U.S and Vietnam. This is a critical problem since with peer-to-peer lending, individual investors face the credit risk rather than financial institutions, which are professionals in dealing with this risk. Because they are at a disadvantage in comparison to the borrower, P2P lenders have a serious problem of information asymmetry. As a result, P2P lending platforms give information about borrowers and the purpose of their loan to potential lenders. Each loan is also given a grade. The empirical study is based on loans' data collected from Lending Club (N = 127844) from January 2016 to June 2017 that are analyzed by using a model-agnostic explainable AI approach, SHAP (SHapley Additive exPlanations). Keywords: peer-to-peer lending, Lending Club, explainable AI, SHAP values, random forest, Grid search.
L'obiettivo della tesi è proporre una valutazione del rischio di credito spiegabile per esaminare le variabili esplicative nella piattaforma di prestito P2P e fare un confronto tra Stati Uniti e Vietnam. Questo è un problema critico poiché con il prestito peer-to-peer, gli investitori individuali affrontano il rischio di credito piuttosto che le istituzioni finanziarie, che sono professionisti nell'affrontare questo rischio. Poiché sono in svantaggio rispetto al mutuatario, i prestatori P2P hanno un serio problema di asimmetria informativa. Di conseguenza, le piattaforme di prestito P2P forniscono informazioni sui mutuatari e lo scopo del loro prestito ai potenziali finanziatori. Ad ogni prestito viene assegnato anche un grado. Lo studio empirico si basa sui dati dei prestiti raccolti dal Lending Club (N = 127844) da Gennaio 2016 a Giugno 2017 che vengono analizzati utilizzando un approccio AI spiegabile indipendente dal modello, SHAP (SHapley Additive exPlanations). Parole chiave: peer-to-peer lending, Lending Club, explainable AI, SHAP values, random forest, Grid search.
Explainable credit scoring in P2P lending: a comparison between the U.S. and Vietnam
DO, THANH THUY
2020/2021
Abstract
The objective of the dissertation is to propose an explainable credit risk assessment to examine the explanatory variables in P2P lending platform and make a comparison between the U.S and Vietnam. This is a critical problem since with peer-to-peer lending, individual investors face the credit risk rather than financial institutions, which are professionals in dealing with this risk. Because they are at a disadvantage in comparison to the borrower, P2P lenders have a serious problem of information asymmetry. As a result, P2P lending platforms give information about borrowers and the purpose of their loan to potential lenders. Each loan is also given a grade. The empirical study is based on loans' data collected from Lending Club (N = 127844) from January 2016 to June 2017 that are analyzed by using a model-agnostic explainable AI approach, SHAP (SHapley Additive exPlanations). Keywords: peer-to-peer lending, Lending Club, explainable AI, SHAP values, random forest, Grid search.È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/987