Introduction to Machine Learning with Applications to Credit Risk<\/b><\/h4>\n<\/div><\/div><\/div><\/div><\/div><\/div>\n\n The use of statistical models to predict banking risk is a standardized practice in the financial sector. Among the most widely used tools are regression models, above all for predicting default and building credit scores. However, the exploitation, generation and availability of data over the last twenty years, together with the development of new quantitative models, have made it possible to use data to obtain better performance in risk prediction. Indeed, the evidence suggests that Machine Learning methods can achieve greater accuracy than traditional models, mainly through more flexible functional estimation. This finding is not trivial, since even small improvements in model performance can translate into significant increases in profitability.<\/span><\/p>\n<\/div>\n<\/div><\/div><\/div><\/div><\/div><\/div>\nFacilitator: Jos\u00e9 Antonio Pellerano<\/b><\/h4>\n<\/div><\/div><\/div><\/div><\/div><\/div>\nMore than 15 years of experience as an Applied Econometrician<\/i><\/p>\n
![\"\"](\"https:\/\/i0.wp.com\/gestionderiesgo.org\/wp-content\/uploads\/2025\/06\/joseantoniopellerano.jpeg?resize=227%2C227&ssl=1\")
<\/span><\/p>\nHe has taken part in various consulting and research projects including collaborations with Nestl\u00e9 Dominicana, the World Bank, the Inter-American Development Bank, the Quito Electric Company, the Dominican Internal Revenue Service, the Vice Presidency of the Republic, the Santo Domingo City Council, the Unified Beneficiary System, Analytica, among others. He has also taught at the undergraduate level in the United States and at the graduate level in the Dominican Republic. <\/span><\/p>\nHe earned his PhD in Economics from Texas A&M University, a master\u2019s in Economics from the Pontificia Universidad Cat\u00f3lica de Chile and a bachelor\u2019s in Economics from the Pontificia Universidad Cat\u00f3lica Madre y Maestra. His most recent research and consulting work focuses on applying experimental and quasi-experimental methods with the main goal of promoting evidence-based decision-making in both the public and private sectors. He has also developed various Data Science initiatives to promote data training and use in the country in a broad sense.<\/span><\/p>\n\n\nProgram Objective<\/b><\/h4>\n
The main objective of this course is for participants to be able to develop their first Machine Learning models applied to credit risk. Beyond that, it seeks to foster participants\u2019 curiosity and empowerment so that they can adapt the knowledge acquired to their institution\u2019s needs. The methodology is based on discussion and the application of code with practical examples (learning by doing). The course format is in person in order to encourage discussion among participants. The course will be taught in \u201cR\u201d and includes an introduction to this programming language, so no prior knowledge is required, although it is recommended<\/span><\/p>\nContent:<\/b><\/h4>\n
Module 1: Introduction to Machine Learning<\/b><\/p>\n\n- Machine Learning, Artificial Intelligence and Data Science<\/li>\n
- Basic concepts of Machine Learning<\/li>\n
- Practical example<\/li>\n<\/ul>\n
Module 2: Introduction to \"R\"<\/b><\/p>\n\n- Data handling and description<\/li>\n
- Data visualization<\/li>\n
- Code with examples<\/li>\n<\/ul>\n
Module 3: Statistical Modeling<\/b><\/p>\n\n- Logistic model<\/li>\n
- Machine Learning Models<\/li>\n
- Important concepts:<\/strong><\/li>\n
- a) Cross Validation<\/i><\/li>\n
- b) Tuning<\/i><\/li>\n
- c) Overfitting<\/i><\/li>\n
- Model performance metrics<\/li>\n
- Applications:<\/strong><\/li>\n
- a) Credit origination model<\/i><\/li>\n
- b) Credit behavior model (Behavioral Scoring)<\/i><\/li>\n<\/ul>\n<\/div>\n<\/div>\n<\/div><\/div><\/div>
The use of statistical models to predict banking risk is a standardized practice in the financial sector. Among the most widely used tools are regression models, above all for predicting default and building credit scores. However, the exploitation, generation and availability of data over the last twenty years, together with the development of new quantitative models, have made it possible to use data to obtain better performance in risk prediction. Indeed, the evidence suggests that Machine Learning methods can achieve greater accuracy than traditional models, mainly through more flexible functional estimation. This finding is not trivial, since even small improvements in model performance can translate into significant increases in profitability.<\/span><\/p>\n<\/div>\n<\/div><\/div><\/div> More than 15 years of experience as an Applied Econometrician<\/i><\/p>\n He has taken part in various consulting and research projects including collaborations with Nestl\u00e9 Dominicana, the World Bank, the Inter-American Development Bank, the Quito Electric Company, the Dominican Internal Revenue Service, the Vice Presidency of the Republic, the Santo Domingo City Council, the Unified Beneficiary System, Analytica, among others. He has also taught at the undergraduate level in the United States and at the graduate level in the Dominican Republic. <\/span><\/p>\n He earned his PhD in Economics from Texas A&M University, a master\u2019s in Economics from the Pontificia Universidad Cat\u00f3lica de Chile and a bachelor\u2019s in Economics from the Pontificia Universidad Cat\u00f3lica Madre y Maestra. His most recent research and consulting work focuses on applying experimental and quasi-experimental methods with the main goal of promoting evidence-based decision-making in both the public and private sectors. He has also developed various Data Science initiatives to promote data training and use in the country in a broad sense.<\/span><\/p>\n The main objective of this course is for participants to be able to develop their first Machine Learning models applied to credit risk. Beyond that, it seeks to foster participants\u2019 curiosity and empowerment so that they can adapt the knowledge acquired to their institution\u2019s needs. The methodology is based on discussion and the application of code with practical examples (learning by doing). The course format is in person in order to encourage discussion among participants. The course will be taught in \u201cR\u201d and includes an introduction to this programming language, so no prior knowledge is required, although it is recommended<\/span><\/p>\n Module 1: Introduction to Machine Learning<\/b><\/p>\n Module 2: Introduction to \"R\"<\/b><\/p>\n Module 3: Statistical Modeling<\/b><\/p>\nFacilitator: Jos\u00e9 Antonio Pellerano<\/b><\/h4>\n<\/div><\/div><\/div>
<\/span><\/p>\nProgram Objective<\/b><\/h4>\n
Content:<\/b><\/h4>\n
\n
\n
\n