In this study a CHAID-based approach to detecting classification accuracy heterogeneity across segments of observations is proposed. This helps to solve some important problems, facing a model-builder: 1. How to automatically detect segments in which the model significantly underperforms? 2. How to incorporate the knowledge about classification accuracy heterogeneity across segments to partition observations in order to achieve better predictive accuracy? The approach was applied to churn data from the UCI Repository of Machine Learning Databases. By splitting the dataset into 4 parts, which are based on the decision tree, and building a separate logistic regression scoring model for each segment we increased the accuracy by more than 7 percentage points on the test sample. Significant increase in recall and precision was also observed. It was shown that different segments may have absolutely different churn predictors. Therefore such a partitioning gives a better insight into factors influencing customer behavior.

Applying CHAID for logistic regression diagnostics and classification accuracy improvement

In the literature, the relation between corporate governance and enterprise value has been extensively studied, and has been found to be positive. However, it is assumed that the impact of improvements in corporate governance practices and enterprise value is homogeneous for any company. We developed a logit model with interactions per country and sector which makes it possible to find the probability of improving the Tobin’s Q when the MILA corporate governance index (MCGI) rises. We found statistically significant evidence that the probabilities differ for the sectors and countries studied.

Corporate Governance and probability of improve Tobin´s Q: A Study Case in Latin America

The process of effective credit risk assessment plays an important role in the financial decision making in Microfinance Institutions (MFIs) as it enables faster credit approval decisions and diminishes the possible risks associated with customers’ repayment defaults. Credit scoring is the most commonly used technique for evaluating the creditworthiness of loans which has gradually begun to find its way into the microfinance field. Many parametric and nonparametric techniques have been adopted by financial institutions to develop accurate credit scoring models. In this study, a credit scoring model is developed for a Tunisian Microfinance Bank by applying fuzzy classifiers where the fuzzy knowledge bases are optimized through differential evolution. Further, the performance of the proposed model is compared to that of the decision tree model. The obtained results reveal that the proposed model consistently gives a better average correct classification rate than the decision tree model. As with the decision tree model, the proposed model can be easily understood by any user and is very useful in the context of credit evaluation process, since it is in ‘if-then’ rule form; unlike decision tree model, the proposed model does not stay in a black box. In the proposed model, the interpretation of independent variables may provide valuable information for bankers and consumers, especially in explaining why credit applications are rejected.

A Credit Scoring Model for Microfinance Bank Based on Fuzzy Classifier Optimized by a Differential Evolution Algorithm

Abstract The objective is to present a proposal for a gender-sensitive risk analysis model using artificial intelligence (AI) within the framework of complex thinking, specifically for impoverished and vulnerable women. This international non-parametric study analyzed business and entrepreneurial activity in a sample of 2,787 women. The methodological design included data analysis, the postulation of a proposed model, and a validation method for the credit risk analysis model. There is a correlation between the level of schooling of impoverished and vulnerable women with the possibility of self-employment and selling a product or service. In the framework of complex thinking, the perception of innovative thinking is related to the level of education and innovative decision-making in professional projects. Women with a higher level of schooling tend to think about their professional projects systematically. Promoting complex thinking involves innovative educational practices to encourage critical, systemic, scientific, and innovative thinking in entrepreneurship and sustainable development. Integrating reasoning for complexity benefits women and contributes to economic and social growth in vulnerable regions. In contrast to other models, our credit risk analysis model uses AI and variables for gender, vulnerability, and complex thinking to detect patterns in women's behaviors and attitudes in the venture start-up process. 

Financial Inclusion of Vulnerable Sectors with a Gender Perspective: Risk Analysis Model with Artificial Intelligence Based on Complex Thinking

This paper provides a literature review on risk scoring models for credit granting in personal banking The methods by Abdou & Pointon (2011), Glennon, Kiefer, Larson, & Choi (2008), and Saavedra-Garcia (2010) are considered The aim is to create a sorting scheme to explain the multiple mathematical and econometrical models used for credit scoring and to produce an up-to-date list supported by scholars and experts in the field

/pdf/modelos-y-metodologias-de-credit-score-para-personas-12d3xvpmiv.pdf

David Esteban Rodríguez-Guevara

Papers

Modelos y metodologías de credit score para personas naturales: una revisión literaria