Reasons for using a neural network for loan risk

[sts-ryan-holton]

@TyMick

Hi, I'm reaching out to query the decision behind building a loan risk network. I'm trying to figure out what made you choose this approach compared to writing a function an essentially assigning weights to each item using if/else statements as apposed to trying to normalise data coming in and building a neural network?

For example, if someone's risk of not paying back is on a scale of 0 to 100, where 100 is unlikely to pay back, why not just create a "points" system assigning a set of points to each field based on the value, like saying if someone's loan amount is super high and their borrowing term is super low to assign a higher weight?

Just trying to uncover this reason