when all winners in every battle is model A (or model B or tie), the Evaluating will get error at the function of "compute_mle_elo"

[Number18-tong]

when all winners in every battle is model A (or model B or tie), the Evaluating will get error at the line 180 of the function "compute_mle_elo" in utils/score_utils.py. It is possible for the situation that all winners are same type, really hope a way to solve the problem.

[RuochenZhao]

Hi, I think this is a problem with the logistic regression function we use. It will need at least two classes to fit the regression: you can see https://stackoverflow.com/questions/63692895/is-there-a-way-to-build-a-logistic-regression-model-even-if-there-is-only-one-cl for this concern.

I'm not sure whether there's a workaround. An immediate solution would be trying to increase the sample size of the debates so that you could get at least two classes?