Meeting your supervisor about a thesis topic

Prof. Clarke: Come in, Jasur. So, have you had any thoughts about your dissertation?

You: Yes, I've got two ideas, but I'm still quite torn between them, to be honest.

Prof. Clarke: That's a good place to be. Go on.

You: The first one is about using machine learning to predict customer churn in telecoms. I've got some industry experience there, so I understand the problem well.

Prof. Clarke: Okay. And the second?

You: The second is more research-based. It's about fairness in credit-scoring models. I find it more interesting, but I'm worried it might be too broad.

Prof. Clarke: What about data? That's often the deciding factor.

You: That's the thing. For the first one, I'd need company data, which might be difficult to get. For the second, there are some public datasets I've found.

Prof. Clarke: Then I'd lean towards the second, but you'd need to narrow it down. Fairness across which groups? Using which methods?

You: Right. Would it make sense to focus on one specific fairness metric, and test it on two or three datasets?

Prof. Clarke: That sounds much more manageable. Could you write a one-page proposal for next time?

You: Sure. How often would you like us to meet, generally?

Prof. Clarke: Every two weeks is usual. Just send me something to read a couple of days before each meeting.

You: Okay, that's clear. So I'll send you the proposal by next Wednesday, and we'll meet on Friday?

Prof. Clarke: Perfect.