I started looking at the code recently and have two short questions/remarks on the odd_ratio_mean function:
The docstring says that the recommended value from @eartigau et al. is 2e-3, but the default value value is 2e-4. Is this intended or should the default value be the recommended one ?
Could it be worth to add the desired precision as a kwarg ? This might speed up the calculation in simple cases where most of the nmax iterations would be useless (I have not run the lbl code on a significant dataset, so I don't know if this would have any impact in the the end, but intuitively it could sometimes help, a warning could also be raised if desired precision is not achieved after nmax iterations).
I started looking at the code recently and have two short questions/remarks on the
odd_ratio_meanfunction:2e-3, but the default value value is2e-4. Is this intended or should the default value be the recommended one ?nmaxiterations would be useless (I have not run the lbl code on a significant dataset, so I don't know if this would have any impact in the the end, but intuitively it could sometimes help, a warning could also be raised if desired precision is not achieved afternmaxiterations).