New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Benchmarks #212
Comments
I had a go at the code you included, and did some comparative benchmarking of my own. The problem is there really is no comparison bewteen what the two algorithms do. More importantly, routing with a single set of weights is rarely useful in any real-world scenarios. Either routes can't represent actual preferences and so are unrealistic, or else distances and/or times are unrealistic. Further complications arise in Plus there's the issue of time penalties, which I'm not opposed to the idea in general here, but if it is to be done, it would need to highlight at least the effects of all of these kinds of differences, which would require a lot of work. Happy to see where this goes .... |
I think
Source: https://github.com/vlarmet/cppRouting#work-with-dual-weighted-network |
UpdateAdding dual weights slows That said, there are two notable things that The other aspect is the flow aggregation / vehible allocation routines. Those are really good, and so well implemented that it's probably not really worth trying to copy them here. I might just start using those routines myself. Whatever happens, I'm happy to keep this issue open for a while to explore a bit further. Thanks for the impetus! |
Thanks Mark glad to hear your take on it and to hear that |
Hi Mark, how would you use the flow agregation routines from cppRouting, based on an existing dodgr graph ? Can I just pass the from / to / time columns to cppRouting's makegraph ? |
@FlxPo Yep, that would work. The results will be different - and maybe very different - from dodgr results, depending on what kind of network weighting you used. In short: cppRouting is fast; dodgr is accurate. |
Closing this; can re-open later if desired |
It could be good to do some reproducible benchmarks.
Opening this issue to track the idea and related code...
The text was updated successfully, but these errors were encountered: