11.2 | Evaluate ML Methods Ethically: Avoid P-Hacking for Reprod. Results | Medical Software Course
from YaleCourses Hosted by YaleCourses
In this lesson, we delve into the critical aspects of evaluating machine learning methods and the vital importance of data integrity. We'll explore fundamental concepts like separate training and testing sets, cross-validation techniques, and the significant challenge of 'p-hacking' that can undermine research reproducibility. Understanding these issues is crucial for developing robust and trustworthy medical software.Uploaded 22-08-25🎯 Learning Objectives• Understand the core principles of evaluating machine learning models.• Differentiate between training and testing data, and the importance of generalization.• Explain K-fold and leave-one-out cross-validation techniques.• Recognize the prevalence and impact of data cleaning in real-world machine learning applications.• Define 'p-hacking' and its implications for the reliability and reproducibility of machine learning research.
Embed this episode
No transcript yet.