Choosing a class for Naive Bayes Classifier

For multiple class in Naive Bayes Classifier, we need to choose the class with the highest posterior probabilities. But what if we have two class with same posterior probabilities? Which class should we take?


Less is more. Stay pure. Stay poor.
do you have sparse data, so it is possible to have two with the 'exact' same value? If you really have two with the same value, I suppose it is a contextual decision - so which seems like the best reference or what have others in your field used as the reference.

It has been a long time since I did NB, but this is for the ratio comparison?
For the data I already estimate the NA. My data has 10 covariates such as Bmi, Gender, Age, type of diabetes. For every covariates, i divided them into groups. For example, bmi we have under weight, normal, overweight and obese. And gender we have male and female. I find the possible combinations for each covariates such as gender female, bmi normal and age young adult. So,to calculate the posterior i need to know the prior prob and the covariates prob, but some of my data which is for bmi underweight, the probability are 0 which makes the posterior for each class become same value which is 0. So i dont know which class should i take if the three classes are 0.