Someone wrote a notebook. It runs. It produces numbers. The question is whether you can tell someone else what it actually does.
Three notebooks. Three different problems. One underlying question: what does the code actually do? Pick one. Open it. Go through it cell by cell.
Three numbers per customer: how recently they bought, how often, and how much. The notebook groups customers by those numbers. The interesting part is what the groups turn out to mean.
Fraud is rare. A model that always predicts "not fraud" scores 99% accuracy. This notebook explains why that is a problem, and what to do about it.
Predict next week's sales from historical data. The simplest useful model, applied to a real retail dataset. Pay attention to how the data is prepared before the model sees it.
Kaggle notebooks are free to view without an account. An account lets you run the code. Running is optional. Reading is not.
Read your chosen notebook from top to bottom. For each cell, ask one question: what is this cell doing, and why does it come here in the sequence?
You will not understand everything. That is fine. The goal is to be able to explain where is the data, what they did to it, what came out.
You will explain your notebook to someone who read a different one. They will explain theirs to you. Neither of you will have seen the other's code before.
What do you think the hardest part of this task will be? What should you focus on?
Pick one cell in the notebook you found genuinely difficult to explain. Write one sentence about what it does. Do not look anything up. Do not ask an AI. Whatever you can say with confidence is what you understood. Bring that sentence to class.