Loan-Status

It is a very common phenomenon for a person to take a loan from financial organizations such as banks, credit unions, and financial institutions. A bank loan is the easiest and most preferable source of taking loan because they provide good loan policies such as less interest rates and longer loan terms. For a variety of purposes banks get a huge number of loan applications every day from people. In many cases people do not repay their loans to banks, therefore, banks suffer tremendous amount of loss every year. Banks have limited assets, and they cannot grant everyone a loan. Banks have to select applicant on the basis of certain criteria. Making a decision on a loan approval has immense risk. In this project, risk factors were reduced by choosing a perfect and safe person so that banks can save lots of time and assets in the selection process of loan applicants. The process of choosing the safe and right person was done by data mining of previous records of the applicants who have applied for a housing loan. On the basis of those records, several classification algorithm techniques such as logistic regression, nearest neighbors, support vector machine, kernel support vector machine, decision tree, random forest, and artificial neural networks were trained for providing the most accurate results. Different machine learning algorithms were compared in this project. The best model found could be used by banks for predicting loan status. After getting significantly accurate results from the trained machine learning model, that model can be applied to predict future applicant’s data whether giving loan to that particular applicant will be safe or not, which is the main objective of this project.