Visit Lending Club (Links to an external site.) which provides data about loan applications it has rejected as well as the performance of loans that it issued. Locate 2018 Q3 data for the Loan Database and the Declined loan data. Locate the Lending Club data dictionary for the loans that were approved and funded. Note all of the data attributes listed in the Excel files (csv) as fields, Which attributes do you think might predict which loans will go delinquent and which will ultimately be fully repaid? How could we test that? Now consider the declined loans data set of LendingClub for Q3 2018. What three items do you believe would be most useful in predicting loan acceptance or rejection? What additional data do you think could be solicited either internally or externally that would help you predict loan acceptance or rejection? If you were in a position to accept or deny a loan application, how might you look differently at this data? Would you be more lenient or stringent? After reviewing your classmate's responses, do you agree or disagree with their position? Did their rationale change your mind? When reviewing the loan data for Lending Club, the attributes that I think might predict if a loan will go delinquent and which will be repaid are a borrows annual income, installment amount, and debt to income ratio. If a borrows annual income is high and their installment payment is lower, chances are the borrow has better cash flow to repay the loan. The debt to income (DTI) also comes into play here. The DTI helps lenders measure the ability of a borrow to repay a loan [ CITATION Con19 \\ 1033 ]. The DTI ratio takes the borrows total debt and divides it by the borrows income. A lower DTI indicates a higher probability that the loan will be repaid, whereas a high DTI indicates the borrow may have problems paying back the loan. This can be tested by looking at a list of delinquent accounts to see if their DTI is higher and also comparing their income and installment payment. I would assume that we would see a correlation between the three. Considering the declined loans data from Lending Club, the three attributes I believe would be most useful in prediction loan acceptance or rejection is risk score, DTI, and employment length. Other stats that would be helpful to predict loan acceptance or rejection would be their past employment. For instances a borrow could have just left their job of 10 plus years for reasons beyond their control such as the company closing. Looking at only their current length of employment could have a negative effect for the borrow in this instance. I think it is important to look at each potential borrow in depth and not just use an algorithm. As such, if I was in a position to accept or deny a loan application, I would look beyond what the numbers are telling me. I would ensure due diligence was completed for each applicant
