Really good job on this course, a huge thank you to all the instructors for providing free access to such a great resource and being available to help on the forums.
Just to provide some context to this feedback: I completed Andrew Ng’s machine learning course on coursera previously so I had some background to the theoretical aspects before this course.
Positives
- Good pace with gentle introduction to concepts with code examples to support everything
- Wrap up exercises are a great touch - I learnt the most by doing these since the questions were a bit vague and needed some more thinking before jumping straight into importing something from sklearn to solve the problem
- Level of support on the forums, the pedagogical team respond quickly to any issues
- Visualisations and explanations in the notebooks are very clear and the motivations behind the code are very clearly explained
Improvements
- I would’ve liked a bit more content on data preprocessing and handling different categories of data. It was touched on at the start with the numerical and categorical data types notebooks, and also later when mentioning how time series data has different cross validation requirements. But since data is sometimes (or most of the time) more important than the model, I think that would be a really interesting future course since most courses focus on the modelling aspect (maybe because its more fun
)
Some people may suggest to include more theory on the models (e.g how linear regression learns coefficients through gradient descent), however I would say that there are plenty of other courses for that (e.g. andrew ngs’ ML course on coursera) and for me I wanted to mainly learn the tools available in scikit learn that would abstract those details away to be more productive.
Again I want to just express my gratitude to the team and I look forward to any other courses you put out there 