Feels like a decision tree is not the kind of algorithm that can solve this problem due to its multidimensional nature. It's very vulnerable to Simpson's parado

Feels like a decision tree is not the kind of algorithm that can solve this problem due to its multidimensional nature. It's very vulnerable to Simpson's paradox. It may be right most of the time if we lump everyone in the same bag, but that is not helpful at all. Also, having the self-evaluated unemployment risk the first deciding factor seems to defeat the purpose.