ACHINEAUTO AI HOME LOAN MODEL REVIEW
It is a known global fact today that most people cannot afford to buy or build a home without acquiring a bank loan. A situation that has left millions of low-income earners homeless all over the world.
This review which was consequent on a machine language modelling, exposes the factors that fosters Banks to reject a home loan request or limit the amount an applicant is entitled to. Few of the factors adopted in this modeling include applicant’s employment status, monthly income, credit history, loan amount etc.
The report predicted that an applicant’s monthly income is a major consideration when it comes to loan approvals. This is because banks understand that it is imperative for you to still have surplus after you might have paid your equated monthly installment, hence, you will be financially handicapped and will not only be at risk of defaulting but might also be a risk to the community by cutting corners to making ends meet.
The Auto AI experiment began with a download of a csv data from Kaggle which was fed into Watson Studio at the instance of machine learning bu. Loan Amount Thousands was chosen to be predicted, i.e the amount that an applicant is entitled to, in case the loan will be approved.
The experiment was run with default data source information while adopting regression prediction type. As presented in the attached image, pipeline 7 ranked number 1. Further examination of the results shows details of the model evaluation measures, model information. Some realignment took place as seeing thereby transforming into new features.
Ultimately, the feature importance which is the most important predictor of the amount of loan approved turns out to be Feature 0 which is the applicant’s and co-applicant’s monthly income. This is by far the most important feature, while other factors are basically to take cognizance of other financial commitment of the applicant (to his/her family), like the marital status which could add up for a financial strength on the other way round (as co-applicant).
Ultimately, the model was deployed and tested using the data of two different applicants ID. The results showed that, while it is predicted that one applicant is entitled to 130, the other applicant is entitled to 245. This implies that it is safe to conclude that their total monthly income is directly proportional to their entitled loan amount.