Skip to content

A deep neural network and a linear regression model for predicting credit scores of individuals using tests instead of history

License

Notifications You must be signed in to change notification settings

soulsyrup/Deep-neural-network-and-linear-regression-for-credit-prediction-of-individuals-using-tests

Repository files navigation



z

Abstract

Credit scores allow individuals and businesses to be able to take loans, many of which stimulate the economy, allowing ideas to be formed into products and services, and giving the opportunity for people to own their own place of residence. The hypothesis of this study was that IQ scores and financial literacy scores are positively correlated with credit scores, and that higher financial literacy confidence scores coupled with low financial literacy scores are negatively correlated to credit scores, thus allowing the possibility to create an algorithm to predict credit scores of people who do not have credit scores. A survey was used to collect the data in this study, specifically IQ scores, financial literacy scores, financial literacy confidence scores, and credit scores. 96 participants, aged 18 and over, were recruited through opportunistic sampling and consented to take part in the survey. Analysis consisted of correlation charts, heatmaps, scatter plots, multivariable regression, to find patterns in the data. The results show that the combination of IQ scores, financial literacy scores, and financial literacy self-confidence scores, is strongly positively correlated to credit scores. A multivariable regression model was created, with an accuracy of 74.19%. An Artificial Neural Network was created, with an accuracy of 65.51%.



SocArXiv Article



https://osf.io/preprints/socarxiv/g4vjt/

syrup, s. (2023, September 29). Alternative Personal Credit Scoring Tests without Financial History, A Novel Method, Credit Needs and Democracy. https://doi.org/10.31235/osf.io/g4vjt

About

A deep neural network and a linear regression model for predicting credit scores of individuals using tests instead of history

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages