Ling, Kim Sia (2023) Modelling of tenant’s behavior information characterizations for credit scoring using logistic regression in Malaysia. Masters thesis, Universiti Tun Hussein Onn Malaysia.

Preview	Text 24p LING KIM SIA.pdf Download (478kB) | Preview
	Text (Copyright Declaration) LING KIM SIA COPYRIGHT DECLARATION.pdf Restricted to Repository staff only Download (335kB) | Request a copy
	Text (Full Text) LING KIM SIA WATERMARK.pdf Restricted to Registered users only Download (2MB) | Request a copy

Abstract

In Malaysia,themortgageapplicationsofthelowincomegroupareusually rejected bybanksandfinancialinstitutionsastheyhavepoorcreditscoresdue to toolittleorevennocredithistory.Therefore,theynormallyrentaproperty, buttheirrentalpaymenthistorydoesnotaccountableinthemortgage applications. Thisstudyaimstoreducethecreditunscorableoflowincome group withlimitedcredithistory.Inthisstudy,thelogisticregressionisapplied to computethecreditscoreoftenantsbasedontheircharacteristics,without relying onthetenant’scredithistory.Thepenalizedmaximumlikelihood estimation withridgeregressionisutilizedtofindtheparametersofthelogistic regressionmodelastheexistingseparationintrainingdata.Theinitial9factors considered affectingtenants’creditscoreweregender,age,maritalstatus, monthly income,householdincome,expense-to-incomeratio,numberof dependents, previousmonthlyrentandnumberofmonthslaterentalpayments. The maritalstatusfactorwasthenremovedfromthelogisticregressionmodel as itisinsignificanttothemodel.Meanwhile,k-foldcross-validationwithGrid Search wasappliedtodeterminetheappropriateregularizationstrengthvalue for maximumlikelihoodestimation.Themainfactorsofthetenant’scredit score arethenumberofmonthslatepayment,gender,expense-to-incomeratio, previousmonthlyrentandage.Besides,thereisnounderfittingoroverfitting in theproposedcreditscoringmodel.Meanwhile,theaccuracyoftheproposed tenant’screditscoringmodelontestingdatais0.90.Lastly,agraphicaluser interfacewasdevelopedfortenant’screditscoring

Actions (login required)

View Item