Hi organizers, and thanks for running this Study Jam — it's a great dataset to learn on.

While exploring the features, I noticed that a couple of the loan-level fields seem to carry information that may reflect what happened after the loan, rather than what's known at disbursement. I wanted to check my understanding rather than assume anything.

Specifically, Total_Amount_to_Repay (and the repay-to-principal ratio derived from it) and Lender_portion_to_be_repaid separate defaults very strongly. Within the same loan_type, defaulted loans show a much higher repayment-to-principal ratio than repaid ones — a pattern that looks more like penalty or outstanding amounts added once a loan has gone bad than a contracted repayment fixed at origination.

A model that uses these reaches a very high F1, but I'm not sure it reflects predicting creditworthiness at the time of lending.

So my question: are these fields intended to be known at loan origination, or would you recommend treating them as leakage and excluding them for a realistic "predict-at-disbursement" setup? Just want to make sure everyone models the intended scenario — the starter notebook includes them, so this applies to all of us equally.

Thanks a lot, and happy to share more detail if useful! BR Andre