urbananalyst / uaengine Goto Github PK

Because the data are all pre-calcualted anyway, so no extra cost to include that as an additional output variable. Weighted by population density, of course.

Following on from updates to m4ra in UrbanAnalyst/m4ra#14

They fail because of path stuff on m4ra, even though those workflows currently all work fine. Those two have now been split into a checks-extra-os workflows file.

Use dodgr::dodgr_dists_categorical() calculation with edges in 3 categories of:

1. Perfect for bicycles (green, parks, separated bicycle ways);
2. Okay for bicycles (bicycle lanes alongside traffic); or
3. Unsuited (in vehicular traffic without bicycle lanes).

Weight those respectively 1, 0.5, and 0, add the categorical weights along all ways from each point out to distances of 5 and 10 km, and convert to proportional distances. Average over all to get metric of local bicycle infrastructure. That will then simply need to be converted to metric of "bicycle unsuitability", as 1 - metric, so that lower values are good like all other UTA stats.

Similar to bicycle infrastructure metric #10, as proportional distances along edges pre-categorised as passing through or next to green or blue space (parks or water).

It will probably be more efficient to restrict the final matrix manipulation step of node -> GTFS -> node to reduce the matrices down to only those components within some limit, say 1.5 times, the maximal distance threshold submitted in the call.

To extract all osm data from a local OSM .bz2 file, or via osmdata.

Hi Mark, just happened across this and sounds amazing! I've just been playing with cppRouting and wondered it would be worth comparing with it in addition to comparisons with valhalla and r5.

https://github.com/AngusJohnson/Clipper2 - old clipper is currently failing on Windows builds anyway, so the update is quite urgent to fix that.

So that uta_interpolate (#1) returns the original graph, not the contracted version.

Very straightforward, and just needs a new initial OSM layer of schools. OSM Wiki has page for "Education features" which suggests reasonable identification of schools only, excluding kindergartens or other facilities as well as universities or institutions of higher education, would be:

• amenity=school or
• school=elementary | primary | secondary

Currently fails because of clipper, so should be fixed by #14. In the meantime, windows checks have been moved to additional check file. Delete this and re-instate windows checks once fixed, then close this issue.

m4ra has now been modified to also output the actual multi-modal duration matrix, and this is now used to return summary statistics on both relative (former form) and absolute travel times. These now need to be summarised as part of the batch_collate_results() and plot_polygons routines, to enable plotting of both relative and absolute values.

Calculation of travel time ratios for every single point is computationally expensive. Equivalent results can likely be gained by calculating values for some subset of all points, and then interpolating between those. This could be done by using a routing algorithm to only track to nearest 3 or so points from each, and to aggregate the distance-weighted ratio values for those points.

Once that's done, the values at each point can then be projected onto the network edges to generate full edge-based visualisations which will be much easier to visually interpret than point-based results.

Now calculated in m4ra via UrbanAnalyst/m4ra#17, so can also be included here.

Remove number of transfers, to leave just travel time times interval. Number of transfers is fundamentally different and, among other things, differs with mode. Including this confounds the combined measure; removing it leaves a clearer measure.