Diseño de un modelo de credit scoring que fomente la inclusión social al sistema financiero haciendo énfasis en las características del bien a financiar mediante técnicas de inteligencia artificial explicables .

Abstract

Nowadays, in Ecuador, to predict payment probabilities, financial institutions continue to use traditional techniques based on statistical models that analyze the capacity of an individual to pay, this even though there is a huge amount of transactional data that could better predict the probability of repayment and encourage financial inclusion, while at the same time preventing the illegal trade of personal data. This collection of data, which is unique to each financial institution, requires Artificial Intelligence techniques for its handling to better take advantage of the available computational capacities. This master thesis designs a Credit Scoring model that offers a better solution to predict the repayment of automotive microcredits, adding to the solution variables such as the characteristics of financed goods, destination of funds, economic activity of the candidate, and the benefits that the acquisition will bring. This work shows the effectiveness of Artificial Intelligence techniques, especially those not called “black boxes”, when additionally, data that is not related to the financial capacity of prospects are used. To achieve this, it proposes that an acquisition and quantification methodology be applied to the prospect’s information during the qualification process and later, during the handling of data. In this project, an artifact that improves the scoring process of applicants of automotive microcredits was designed