TODOs for the tutorial

[asmeurer]

Here are some known TODOs from my end. To be discussed at our weekly meeting.

I have checked in a lot of example notebooks from the other notebook repos, because Jupyter lab makes it hard to copy paste cells unless they are in the same directory. But all notebooks that are definitely part of the tutorial are linked from index.ipynb. Use that as the baseline.

INTRO

• We have the start of an "intro to RAPIDS" notebook, but it needs to be fleshed out more. It doesn't have much on GPU basics.

cuDF

• The intro to cudf notebook is ready for review from my end. It may need to have some stuff removed from it.
• The intro to cudf UDFs notebook is also ready for review.
• Do we want any other notebooks for cuDF?

cuGraph

• The intro to cuGraph notebook is ready for review. There isn't much there, relative to intro to cuDF, but cuGraph is also a smaller library. Are any cuGraph basics missing?
• What cuGraph algorithms do we want to show? There are example notebooks for each of them, all using the karate dataset.
• Do we want a cuGraph notebook with a larger dataset? If so, this will need to be written.

cuML

• The intro to cuML/scikit-learn notebook shows a very simple linear regression on a small (plotable) dataset. Is this OK?
• The intro to cuML/scikit-learn notebook needs an exercise
• What algorithms do we want to show? Are linear regression, PCA, and UMAP OK?
• What dataset(s) should we use? The example notebooks show mortgage or fashion, so if we want something else (perhaps something larger to show off performance), they will need to be written. Perhaps NYC Taxi?

OTHER

• GPU containers.

Note that the solutions to the exercises don't render on GitHub (it doesn't support the <details> tag). Look at them on nbviewer or locally.