What is the difference between the discrete choice method and the generalised cost method in transport planning?

I have been using the generalised cost method in an agent based model. The generalised cost method enables each individual agent to make a decision on what mode of transport it will take given a range of options from A to B. The monetary and non-monetary aspects of the journey can be summated with different valuations of the non-monetary aspects dependent on the agent and the mode (waiting is weighted 2* travelling and so on).

Thus, the generalised cost function can be defined as:

g = f(a) + t(b)

Where

g = generalised cost
f = financial cost
t = time cost
a = Value attribution of financial cost (by socioeconomic group)
b = Value attribution of time cost (by socioeconomic group)


I have temporally dynamic time cost (t) values that has enabled me to display how congestion on the roads and dynamic timetables can result in different decisions being made for exactly the same journey but at different times.

I was intrigued by Kenneth Train's Discrete Choice Method work here.

Discrete choice models attempt to predict choices between two or more discrete alternatives. They statistically relate the choice made by each person to the attributes of the person and the output the probability of the agent making each choice in a set of alternatives.

I understand the outputs are different. The generalised cost method will result in the selection of the option with the lowest generalised cost. The Discrete choice method will result in an output of the probability per each choice.

What are the key differences between the methods?