Please use this identifier to cite or link to this item: http://arks.princeton.edu/ark:/88435/dsp019019s521q
DC FieldValueLanguage
dc.contributor.authorMathews, Simran-
dc.date.accessioned2018-08-20T13:05:50Z-
dc.date.available2018-08-20T13:05:50Z-
dc.date.created2018-04-16-
dc.date.issued2018-08-20-
dc.identifier.urihttp://arks.princeton.edu/ark:/88435/dsp019019s521q-
dc.description.abstractLending Club, a micro-credit organization, provides years of transaction history that allow us to build machine learning classification algorithms in order to improve microfinance operations. These predictive models for microfinance allows us to delve into a relatively new application for machine learning. By optimizing how we better predict default rates, we allow for microfinance lenders to better understand their risk and return portfolios. This objective of this paper is bi-fold: 1.) to optimize the loan portfolio with regard to determining whom and how much to fund in any given transaction and 2.) to determine the optimal interest rate for each transaction through machine learning and deep learning built off extensive literature review. The best prediction rates through our deep learning models had around 90 percent accuracy while KNN was able to achieve around 92 percent accuracy. Sampling machine learning scholarships as applied to microfinance allowed me to form an ensemble of methodologies that preformed with better results than any single model from literature. This research and these results contribute toward highlighting the usefulness of applying machine learning to non-profit and microfinance sectors.en_US
dc.format.mimetypeapplication/pdf-
dc.language.isoenen_US
dc.titleLeave Me a Loan: Machine Learning Strategies for Loan Portfolio Managementen_US
dc.typePrinceton University Senior Theses-
pu.date.classyear2018en_US
pu.departmentOperations Research and Financial Engineeringen_US
pu.pdf.coverpageSeniorThesisCoverPage-
pu.contributor.authorid960952924-
pu.certificateEngineering and Management Systems Programen_US
Appears in Collections:Operations Research and Financial Engineering, 2000-2020

Files in This Item:
File Description SizeFormat