A variant on the shortest path problem - we have multiple employees with access to underground stations around London. What would be the best way of choosing a new office location given the starting points of the employees with the primary goal to keep average transit time using public transport to the new location a minimum? And has anybody done this already? (Mapumental have an open service but it seems to be the wrong way round for my purposes<i>). I've not been able to find anything but perhaps I've been asking the wrong question. the ability to weight the importance of various key staff would be nice as well. Of course, we might well end up just schlepping over to Silicon roundabout.<p></i>edited to clarify that Mapumentals service is not wrong, just the way I want to use it
It's a problem which is common in the transport sector to decide on where to build warehouses. Unfortunately I don't have recommendations that are open to use, most seem to be custom in-house solutions.<p>A big question is what you are optimising for? If it's average time, it could be that it is far for everyone. Or are you employees living in clusters that can carry a heigher weight? You already talk about different staff carrying a different weight. etc. etc.<p>It's typically not even easy to properly define the problem :-)
Can't you run Mapumental for each home and then combine the results somehow to find the best location?<p>Actually, do they already support this? They say "Mapumental handles multiple start or destination points too."<p>I was thinking just overlaying the maps as semi-transparent PNGs would reinforce the color gradient of the nearest places.
I would probably start by asking your staff their nearest 2 underground stations. Upload the spreadsheet to something like ZeeMaps to pin all the locations to a map and getting a visual idea of where clusters appear. You could even do this on your office wall.<p>However is the best office one that has the shortest journey? Surely neighbourhood, space, cost & local facilities are equally important.
There are probably a fintite number of points that are your best guess and near a station. Use station locations inside a radius.<p>Then run them all through mapumental to get distances. Weighting can be done in code or a spreadsheet.<p>Or implement a hill climbing algorithm using mapumental and the weights to make your cost function.