I'm performing a logistic regression on a dataset. My dependent variable is a dummy that indicates whether a company is part of a cartel. In my dataset I have 1242 companies: 113 (9.1%) are part of cartels, while 1129 (90.9%) are not. I've been told that it's quite hard to predict 9.1% of cases, but it should be more balanced. However, I cannot do much because that's my dependent and I do not have other companies in the cartels... Do you have suggestion on this?

The question is different from that assumed to be a duplicate, because it's a different database with different percentages of cases and from that answer I'm not able to understand whether in my case the same considerations would apply.


Browse other questions tagged or ask your own question.