Feedback for 1.7: Hyperparameters

Clarity: 4/5
Quality: 4/5
Relevance: 5/5

Errors:
Suggestions:

– sponsoring…

– When you discuss the values on a particular decision tree, you refer to “the left number” being the number of cases in class 0 and “the right number” for class 1. Show it on the slide, or mention that it is the field labeled as “Value”.

– Slide “Vul mij in” at 10:00

– You mention an improved recall after hyperparameter optimization, but don’t discuss strategies on how to target improved recall. How could we expect that changing the hyperparameters like we did, we would end up with improved recall? Because we used accuracy / Matthews correlation as metrics to decide on the hyperparameters.

Location: AI for Materials Industry » Case 1: Defect detection on steel plates using tabular data » 1.7: Hyperparameters

Clarity: 4/5
Quality: 4/5
Relevance: 5/5

Errors:
Suggestions:

– sponsoring…

– When you discuss the values on a particular decision tree, you refer to “the left number” being the number of cases in class 0 and “the right number” for class 1. Show it on the slide, or mention that it is the field labeled as “Value”.

– Slide “Vul mij in” at 10:00

– You mention an improved recall after hyperparameter optimization, but don’t discuss strategies on how to target improved recall. How could we expect that changing the hyperparameters like we did, we would end up with improved recall? Because we used accuracy / Matthews correlation as metrics to decide on the hyperparameters.

  • Actions:

Leave a Reply

Your email address will not be published. Required fields are marked *