Curriculum

  /  

Descriptions

MLDS 490: Interpretable Machine Learning for Finance

Description

The course emphasizes interpretable machine learning techniques and their applications in the financial services industry. Students will develop machine learning models, explain model predictions, and build stakeholder confidence through transparent model outputs. Financial applications include client acquisition, credit underwriting, expected credit loss estimation, and fraud detection. Through hands-on exercises with industry-relevant datasets and open-source tools, students will design interpretable models that drive strategy and mitigate risks.

Learning Objectives
•Interpret, visualize, and effectively communicate machine learning model results to technical and non-technical stakeholders.
•Develop prescreen models to identify and target prospective clients, ensuring interpretability for actionable marketing strategies.
•Build interpretable scorecard models to support loan approval decisions and generate adverse action codes for rejected applicants.
•Estimate expected credit losses with models that provide clear explanations aligned with U.S. accounting standards (CECL).
•Design interpretable models to monitor and detect fraudulent transactions in highly imbalanced datasets.