Introduction
Discrete choice models (DCMs) aim at predicting which alternative a person will choose. The models share similarities with logistic regression although with some fundamental differences in assumptions about the distribution of error terms.
The theory of DCMs has its roots in the random utility theory and assumption that a rational person will always choose an option that will maximize the utility a person gets from choosing such an option.
For example, let's assume that you are to choose between two independent alternatives (that is, such alternatives that do not share any common characteristics); for the sake of this example, we will consider choosing between biking and driving a car to work. It costs nothing to bike to work (we are, of course, assuming that you already have a bike and we do not count the energy that you burn while biking), but the cost of driving a car to work would be $3. However, it takes roughly 45 minutes to get to work on a bike while the same trip using...