|
This paper real estate secured loan one from A.D. 2000 to 2003 of specialized bank of guaranteed loan of house as the target , the situation divides the family in breach into the normal family and exceeds the time limit to stop the paying of the debtor, and with the sex, age, marital status, academic credentials, mode of repayment, the rates of total debt and annual income, room age , area of the house , security position, grant the loan to count , whether the special project grants the loan, while granting the loan, interest rate, amount of money of the loan, whether borrower and loan have geo- relations to walk, whether borrower and security have geo- relations, job, whether related guarantor has, annual income, annual surplus , always in debt, the collateral is owned by me, the collateral is owned by spouse, walk storehouse and security have geo- relation , market price , valuation , day house and apartment building completely , for purchase house and research parameter of checking and approve the authority etc. carry on the discussion, and carry on Logistic Regression Model analysis of the parameter screened . This research paper passes all primitive parameters square independent character examination (P value <0.05 ) and parameter chosen after alternate analysis are established as Logistic Regression Model (1 ), is it is it study aforesaid research parameter that choose parameter too strict and lose kind parameter to choose to avoid another, aforesaid all parameter that choose, so long as value, examination P of square independent character, <0.25 research parameter include in , and the parameter after screening with Stepwise is established as Logistic Regression Model (2 ), through comparing Logistic Regression Model (1)And the model of Logistic Regression Model (2 ) fits excellent degree assay, predict accuracy and predict ability wholly, among them every index of Logistic Regression Model (2 ) is more excellent than Logistic Regression Model (1 ) , among them the whole prediction ability of Logistic Regression Model (2 ) is up to 94.1% 93.7% slightly higher than Logistic Regression Model (1 )s, so regard Logistic Regression Model (2 ) as the risk of making loans finally and assess the way, and assess the way and examine 35 cases of guaranteed loan family of house of A.D. 2004 with the loan risk of this Logistic Regression Model (2 ), predict that the ability rate is up to 75%, can offer the bank to judge the basis effectively completely.
|