mazba-ahamad / 5-Model-Risk-Management-Evaluation-Frameworks Public

Notifications You must be signed in to change notification settings
Fork 0
Star 0

Human decision scientists should consider five critical aspects when evaluating ML models.

0 stars 0 forks Branches Tags Activity

Name		Name	Last commit message	Last commit date
Latest commit History 11 Commits
Appropriate Model Selection		Appropriate Model Selection
LICENSE		LICENSE
README.md		README.md

Repository files navigation

Model Risk Management (MRM): Five Critical Human Interventions for ML Model Evaluations

Decision scientists should consider five critical aspects when evaluating ML models of MRM projects.

i: ML-based feature reduction approaches are not always suitable for business strategy development.

ii: Use of macroeconomic, policy, and shock variables in ML model as "Control Vars."

iii: Less important features within multilevel modeling framework may be critical to increase model performance in production.

iv: Detrending is critical when seasonal anomalies create a linear trend in time-series data.

v: Use of statistical and ML-based models simultaneously to avoid modeling failures.