Dear students,
The grade for Milestone 1 are now released on Moodle. If you did not get the same grade as the rest of your group, or got a very unexpected grade, please contact us by e-mail.
Generally speaking, the submissions were good! We have nonetheless seen a few recurrent mistakes, or things we’d like to see in MS2:
- Data splitting should be done before normalization as we are supposed to normalize only using the training data mean/std.
- If you add a bias term, it should be done after the normalization.
- Do not forget to compare your final models and methods on the test set.
- For the report, do not talk about the specifics of your code and implementation (such as describing your variables, etc.). Instead, give a high-level overview of what you do.
- Plots and/or tables should be used to report results (e.g., when studying a hyper-parameter effect on performance) in your reports.
Best,
Nicolas