Create dataset

[blahami2]

• Generate random node pairs
• Pick pairs in order to get a full range of routes (short, medium, long, ...)

[blahami2]

• Optimize "random" picks
  • Prefer (if needed) ones with the largest geospatial distance (use pythagoras for performance reasons)
  • Impl: sort n^2 nodes by distance, pick randomly with preference for higher values

[blahami2]

Optimization

• Heavily optimized
  • Implemented Geospatial distance heuristic

Procedure

1. Randomly pick N nodes
2. Create N*N pairs
3. Sort them by distance
4. For each interval repeat max K times or until found
   1. Calculate actual route distance
   2. Compare to interval min/max distance
   3. Adjust if not inside the interval (direction +/-), repeat
   4. Save otherwise