How do you produce a hierarchy least cost path and a non-hierarchy least cost path?

[TasBarsdell1]

Hello I am wanting to identify least cost paths by bicycle from census area centroids to a college campus. My purpose is to find which routes/segments are used the the most by cyclists for two outputs. A 'potential' cycling traffic output not using a hierarchy and a 'real' cycling traffic output to show what the likely segments are using the hierarchy attribute:

Using a network data set comprising of on-road cycle paths (hierarchy value 2), off-road cycle paths (hierarchy value 1) and a road network (hierarchy value 3).

Using the network analyst - Closest facility tool to show the least cost path for each census area centroid to the college campus.

Export the solved routes as a layer to then intersect with the original network data set to calculate the traffic counts for each segment to identify which segments are the most popular.

Then repeat analysis using hierarchy and to note the difference in least cost paths produced and the different traffic counts of segments for each output to suggest which segments/roads should have more attention to make more cycle friendly.

Hope I have got my aim across clearly.

I have tried these steps and have produced the 'potential' output easy enough, however when introducing the hierarchy into the analysis I either only get a 'potential' output or a 'real' output. For example if I create a network dataset then use Hierarchy with ranges 1-3 and use the solve for closest facility I get an output that only uses hierarchy levels 1 (off-road) and 2 (on-road) but if I turn off the hierarchy I get the same results. Likewise with I don't use the hierarchy I get the 'potential' output but if I try to use hierarchy it only produces the same 'potential' output excluding the use of hierarchies.

[MelindaMorang · Accepted solution]

Hello, Tas.  I don't think that Hierarchy is really the right tool for the job you are trying to do.  Hierarchy is meant to help very quickly solve long-distance routes on large networks, routes that would take a really long time to solve with an exact solution.  The hope is that the route computed using Hierarchy is the same or very similar to the exact route computed with Hierarchy turned off, just much faster to compute.  Thus, it is not surprising that you got the same solution when you toggled your Hierarchy value.  Hierarchy is meant to decrease computation time, not to control which roads get used in your analysis.

I think you might get closer to what you want by using restriction attributes based on road class (street, off-road cycle paths, and on-road cycle paths).  You can create "soft restrictions" that have a value of "Preferred -[high/medium/low]" which serves to make the route prefer to use those road segments over others.  So you could construct your restriction attribute so that off-street cycle paths are preferred a lot, and on-street cycle paths are preferred a little less (or whatever you think is appropriate) and regular streets are just normal.  Solve the analysis with the restriction turned off for a normal analysis and then turned on to mimic people's preferred behavior.  Here is some information about Restriction attributes: Understanding network attributes—Help | ArcGIS for Desktop

Finally, instead of computing the route and then intersecting it with the underlying street features (which may be imperfect in the results returned), instead, use the Copy Traversed Source Features tool (Copy Traversed Source Features—ArcGIS Pro | ArcGIS for Desktop ).  This returns a feature class of the network edges that were used in the solved route, which is exactly what you're looking for.

[TasBarsdell1]

Hello Melinda

Thank you. What you suggested sounds exactly what I was looking for. I have had a quick play around with using restrictions from the Arc tutorials and will have a go using these for my network tonight. Hopefully all goes perfectly.

The second suggestion looks like it will be a much easier and successful process also.

[TasBarsdell1]

Thanks Melinda, have run the process on my network data set and all is working well. Producing 'Potential' and 'Real' least cost paths and traversed arcs. Will just need to do some table manipulation and summary tables to show the comparisons. Thanks again.