I have the task of creating a map in which the objective would be to help determine a site for a conference. I have the following data:

Addresses of possible conference attendees (I have already geocoded them). 6 possible city sites in the Northeast.

I have the Spatial Analyst extension but not network analyst.

I think a major factor in determining a site would be proximity to the city site and keeping driving time to a minimum for as many people as possible, say less than 3 hours or so or 150 miles).

How can I make a heat map as a general overview of clusters? Answered below I used the kernal density analysis.

How can I figure out the best site based on least driving distance without using network analysis? Answered below.


This may suit your needs:

  1. Create raster layer of your roads (coarse scale is fine).
  2. Calculate all cells in raster layer to equal 1.
  3. Set a definition query on your potential sites so only the first site is displayed.
  4. Calculate the Cost Distance for all roads to that site.
  5. Go back to step 3 and set query for next site. Repeat for all 6 sites. This could be automated with Python, but 6 sites is manageable.
  6. You'll have 6 cost maps, like below, for your area. enter image description here
    1. For each cost map, Extract Values to Points, for your addresses. You'll have 6 point layers, with a distance for each point to the potential site.
    2. Run Summary Statistics on each extracted point layer, calculating the mean value for each site. The lowest mean value is the site where, on average, your attendees will drive the least.

Depending on the distribution of your points, you could do something relatively simple, such as using the Minimum Bounding Geometry tool to create a bounding box. Ideally you would use a convex hull as an option, but depending on your license that might not be available.

In the following screenshot I have, for example, I used rectangle by area:

enter image description here

(assuming the point sare the location of the potential attendees)

Now, all you have to do is finding the center of that bounding box.

You could either use the Feature To Point tool to convert the polygon into a point (which would return the center), or you could calculate it using the Field Calculator. Either way you would have your central point, which you can now use to see which city it is closest to.

Of course this method depends on your data. If you have some extreme outliers, or if your points form heavy clusters, the central point that is returned might be nonsensical.

EDIT: To generate the heat map, since you have Spatial Analyst you could use Kernel Density. But attention: since your extent is quite large it might take a very, very long time to run, depending on what pixel size you will choose for the output. So I would recommend saving everythin else before you test this tool ;-)

  • Hi. Sorry yes I have about 38,000 addresses across the U.S. They are clustered. Lots of points in the NE, SE, and California area. Since this is a conference in the NE, (we have midwest and west conferences too) people in the eastern US will most likely attend. – Andrew Feb 9 '15 at 15:43
  • OK; then my option one will not work well! Check out my edit about the heat map. You could also add a screenshot of your data, so it might be easier for some to come up with a good solution. – BritishSteel Feb 9 '15 at 16:33
  • Hi thanks. I tried the kernal density. I used the default pixel size and it did not take long at all. Maybe I'll decrease the pixel size and compare them. I think this will be useful. – Andrew Feb 9 '15 at 17:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.