In case of the simple multinomial logit model, alternatives will be listed and their percent of choice is collected in data collection.

My question is there are modes which will be less selected and will have less amount in the count. what percent of mode choice count is viable for modeling cases? and why? is it possible and correct to use those modes which have the minimum count of choice from the data?

Any help is appreciated.

More Mollalign Yizengaw Mebrat's questions See All
Similar questions and discussions