Evaluating models without test data
Practical AI
00:00
Challenges in Machine Learning Model Evaluation
This chapter explores the difficulties in assessing machine learning models without conventional test sets, particularly in fields like retail and finance. It emphasizes the need for innovative evaluation methods that incorporate human judgment, while also addressing the complexities of model training, deployment, and monitoring. The discussion highlights the implications of data quality, feature selection, and the use of tools like Weight Watcher in ensuring effective model performance.
Transcript
Play full episode
