Hi,
For those who like to reproduce the notebooks locally, we need to copy the folder “helpers” from python_scripts/helpers (available on the GitHub) in the local folder of the notebook, so that we can visualize the two classes separation with the decision boundary.
More details about that here.
Hi there, out of interest what is your main motivation for creating a local notebook?
Note that if you want to run the notebooks locally, I think the most convenient way is to follow the instructions here
If you follow the instructions at the end you will get an index file that looks like this:
and should make it easier to run the notebooks locally while still following the structure of the MOOC.
I like to be able to reproduce locally, in my own coding environment. The main motivation is that I can detect possible hardware (e.g. exercise M4.04) or environment issues. I may also want to try different approaches, with other frameworks to benchmark. And besides this, I also have my own preferences, for example JupyterLab over Jupyter.
I think the suggestion that you give for the local notebook conda environment is nice for learning, but if I want to reuse some of the code snippets in another environment, I might have some issues.
OK thanks for your answer!
Note that the helpers.plotting module is only used to draw decision boundary plots in the MOOC and that in the long term it will be added to scikit-learn. If you want more details you can look at the ongoing Pull Request here
