I've a rather simple question for a school project. We're developing a GA solution for the following problem:
Chromosome: A location with lat-lon coords. There are two types of locations - up to 15 waypoints from user input, and a dataset of about 3-400 stations.
Gene: A route consisting of all waypoints (incl. a fixed start and end) + 1 station.
Fitness function: Shortest path.
Stop condition: Run duration - configurable, default 3 seconds.
We're discussing two possible implementations:
- A problem set of all waypoints and all stations, kinda like soccer team assignment or nurse rostering design. Run GA once on all of that.
- A problem set of all waypoints and one station, do TSP. Run GA for number-of-stations-in-dataset iterations.
Which is better, in terms of design, efficiency, and performance?