Closest facility discrepancies

39
2
10-18-2019 04:13 PM
Highlighted
by
New Contributor

Hello!

I've run into an issue using the network analyst and I'm hoping others may have some suggestions.

Last year I used the network analyst closest facility tool to identify the supermarket located in closest proximity to each participant in a dataset. I recently decided to run a slightly different analysis and started by rerunning the closest facility tool using the exact same data. However, when I compared the results, 11 out of 197 observations were inconsistent (i.e. different supermarkets were identified as the closest facility).

 

I put the inconsistent results into Google Maps just to check out what was going on and the discrepant supermarkets appear to be about equidistant from the given participant’s home, so it doesn’t appear that there are any major errors in how the closest facility is being determined, just minor discrepancies. I've attached a screenshot where you can see the home, closest facility, and route highlighted on the left, and the previously identified closest facility highlighted on the right. 

I am now working at a different institution than when I ran the first analysis and I'm wondering if the network datasets are slightly different? Both institutions use the pre-built SDC street networks provided by ESRI, but maybe they’re from different years? Or maybe I am specifying slightly different options when I run the analysis? I only recall using the default options previously, which is what I am also doing now, but it’s possible that I specified a slightly different option before. 

 

Has anyone encountered this? Any thoughts are greatly appreciated. Thank you.

Reply
0 Kudos
2 Replies
Highlighted
Esri Regular Contributor

Different Network datasets (SDC over the years) can produce different results as new roads are built or existing may change (re-align, or become one-way, or new turn restrictions have come in, etc.). Also if you are using a Start time to solve the closest facility then solving on a different day/time can change the results.

In your case, for the 11 that are different, you can solve for two closest and see if the second is what you were getting before and what is the difference between them.

In general, if you have solved a closest facility once and there are no changes to the data or analysis settings then you will get the same answer if you solve it again on another day.

Jay Sandhu

Reply
0 Kudos
Highlighted
by
New Contributor

Hi Jay,

Thank you for your reply. It sounds like there are probably some differences between the two network datasets. When I solve for the two closest I get most of the same facilities that I got before and the differences between the two are really small. (There were a few where I needed to solve for the three closest, but the differences were still small).

Thanks,

Ilana