Service Area Analysis results is confusing (long post)

[abdinafan]

I apologize if I'm doing this incorrectly since this is my first time posting here.

I'm trying to make a service area of public facilities in ArcGIS Pro like this picture, with the same cutoff of 1.000 m, 2.000 m, and 3.000 m. I'm using a 'local' network dataset that I made from a road network .shp

Service Area Example

Since the information needed for costs, restrictions, and travel mode in the road network attribute table is sadly very limited, this is the cost (pic 1) and travel mode (pic 2) settings that I came up with

Cost(s): Distance

 Travel Mode Settings

But the result came out like this and the cutoff is nowhere near what I intended. The service area of some of the facilities are also very small, as you can see the two points at the bottom part seems to not have service area if you see it in this scale (about 1:80.000), but if you zoom to the feature it has a service area

Service Analysis Result 

I Googled and tried several things (like changing the cutoff to see if it came out right, input an equation on Distance costs evaluator's value to convert Shape_Length from meter to kilometer, Force Full Builds option checked/unchecked, even Typology to check any unconnected lines) but none seem to work.

My suspicion is I failed to identify the root cause(s) correctly or I can't formulate the right keywords that match my problem. However, I have several hypotheses as what could be the root cause(s)

1. The 'quality' of data in the road network .shp is bad and sadly it cannot be used for the analysis

2. The facilities that have very small service area isn't directly connected to a road network (it's kind of 'in the middle' a polygon), so maybe I have to shift the position a little so it connects to a road nearby?

3. I created a Topology with 'Must Not Have Dangle' rules to see if any of the road is disconnected (some of the Google results suggest that this maybe the cause), but as far as I can see most of the errors identified (9.437 errors out of 22.687 entry) are dead ends or cul de sacs. I've read that it is possible to make exceptions for dead ends and cul de sacs but I can't find how to do it anywhere.

Identified Typology Error

So, what is the common root causes for this type of problem? And what steps and/or solutions that I can do to fix this?

[MelindaMorang · Accepted solution]

I think you're on the right track with your testing so far.  To me, it looks like it's probably some sort of connectivity problem.  It might be a topology issues  (gaps, dangles, poorly connected streets).  Or it might just be something you can fix easily by adjusting the network dataset's connectivity policy from End Point to Any Vertex.

If your street features are split at every intersection (they intersect at endpoints of features), then End Point connectivity (the default) is appropriate.  However, if your street features are long and intersect with many other features at junctions instead of endpoints, then you should instead use Any Vertex connectivity.

Here's some documentation about connectivity: https://pro.arcgis.com/en/pro-app/latest/help/analysis/networks/understanding-connectivity.htm

If that doesn't fix it, you might gain some insights by working through this set of Tasks intended to help local governments (mostly in the US) create network datasets for their cities using street centerlines: https://www.arcgis.com/home/item.html?id=9e321cf94d2b43b991cc5109189b2c26.  I'm not sure if your input data has all the necessary attributes, but you might still learn something from the Tasks, since the steps guide you through some data quality, validation, and editing workflows.

[abdinafan]

Thank you for the answer!

My problem is solved now by changing the dataset's connectivity policy to Any Vertex. I understand now that my road network .shp consist many road sections that made up by several parts, the default connectivity policy (End Point) is 'confusing' the algorithm