Mortgage Loan Data Exploration with Non-parametric Statistical and Machine Learning Perspectives

This paper addresses the difficult problem of housing lending, specifically the question of Mexicans living in the United States asking for a mortgage in Mexico and determining the rate that can be received and under what conditions. This comprehensive research approach starts with collecting data covering detailed information about the applicants, such as their income, credit history, work experience, and current debts, among other vital aspects. A rigorous process of data analysis, processing, and cleaning was carried out, including validation of the completeness and consistency of the information, elimination of outliers or missing values, imputation of missing data, and normalization of variables. In addition, the quantitative feature selection method is employed to identify the most relevant variables to predict the authorized rate for the applicant. This work highlights the incorporation of Topological Data Analysis as an auxiliary qualitative method, which enriches decision-making related to authorized rates. Various machine-learning techniques, such as Linear, Principal Components Analysis, Random Forest, and Neural Networks regression methods, are implemented in the modeling phase. This work presents a comprehensive and multidisciplinary approach to address the challenge of housing credit granting through authorized rates, combining quantitative and qualitative methods to improve decision-making and offering a valuable contribution to the field of credit evaluation.