6 Replies Latest reply on Dec 23, 2013 12:22 PM by rfairhur24

    Feature to Point questions

    goh_raj
      Please see attachment.
      1) Did you guys change/improve the results when running feature to point? See comparison between 9.3(9.3.1) and 10.
      2) I came across bug NIM047493: Requesting that points created by "Feature to Point" using the Inside option be created as non-centroids only if the centroid does not fall inside the polygon. This doesn't seem fixed in 10 Pre release (See circled items on attached image). Will we see the fix in 10 final?

      Thanks.
        • Re: Feature to Point questions
          dlee-esristaff
          Hello Goh_Raj,

          Answer 1): Yes, there has been an enhancement for the INSIDE option of this GP tool in ArcGIS 10; therefore the difference between 9.3 and 10 results are expected.

          Answer 2): The request NIM047493 has been "rejected" because even a centroid is inside the polygon, it may still be too close to the boundary, if not on the boundary. The improvement in version 10 is to avoid point locations too close to polygon boundaries. Your examples showed the improved results, hopefully more satisfactory overall than before. The points are not necessarily at centroids.

          Thanks,
          Dan - in GeoProcessing
          • Re: Feature to Point questions
            goh_raj
            Hello Goh_Raj,

            Answer 1): Yes, there has been an enhancement for the INSIDE option of this GP tool in ArcGIS 10; therefore the difference between 9.3 and 10 results are expected.

            Answer 2): The request NIM047493 has been "rejected" because even a centroid is inside the polygon, it may still be too close to the boundary, if not on the boundary. The improvement in version 10 is to avoid point locations too close to polygon boundaries. Your examples showed the improved results, hopefully more satisfactory overall than before. The points are not necessarily at centroids.

            Thanks,
            Dan - in GeoProcessing


            Hello Dan,

            Thank you for your quick response.  Yes, I do see the improvements.  I have to generate a list of x and y coordinates for our parcels every year (we have over 770,000 parcels) and I thought to use 10 this time for testing and maybe even for production.  I know that I was a little alarmed once I saw the results, because the output was so different that I started wondering which version produced the most accurate results.  Looking at the comparison, yes the results from using 10 HAVE improved and I guess I just wanted someone to tell me that it has.  It is slightly unusual that there are areas (like the ones circled in purple) where the points shifted instead of stayed right at the centroid. When running feature to point with �??inside�?� unchecked, the points in these areas didn�??t fall outside of the polygon to begin with, so I was trying to understand the reason behind the shift.  Normally, I would think running feature to point with the "inside" option checked, all points should stay at the true centroid unless it fell outside of the polygon, then in these cases the point can be placed inside or closest to the center.  In some of these areas, like the polygon located at the 3 o�?? clock location, it does not have the true centroid located that close to the boundaries and I don't think it needed that shift.  This could also be true for the polygons at the 4 and 10 o�??clock position too. However, I can see the need for a shift for the parcel at the 11:30 position, since that is a very small polygon.

            By talking to support, I have developed a process to capture as many single points at the true centroid of the polygon as possible.  This involves running feature to point with �??inside�?� unchecked to capture the true centroid.  Then I selected all the polygons without points and exported these polygons as a separate layer (to run feature to point with the �??inside�?� option checked on).  In turn, I would remove these same polygons from my initial layer so that I can rerun my process to reselect polygons without points and append these results to my second parcel layer. I would rerun this process a few times because running feature to point at default causes some of the �??centroids�?� from irregularly shaped parcels to fall on neighboring parcels and sometimes I cannot get a true count of parcels that do not have their own points inside.  I will probably try to automate this process too.  Yes, I appreciate the improvement from 9.3.1 as it is much better.  Thank you for your explanation.

            I do have an idea and I do understand the need to avoid placing the points too close to the boundaries.  Can I suggest an enhancement to give the user an option to define a buffer/distance first when running the feature to point operation?  For example, when running feature to point with the �??inside�?� option checked on, all points should be generated and placed at the true centroid of the polygon, but if the point is ____ distance away from a boundary, then the point can be shifted to another location. If the point is totally outside of the polygon, it would use the same placement method as before.  Just an idea, thank you for listening.
            • Re: Feature to Point questions
              dlee-esristaff
              Hi Goh_Raj,

              Thank you for carefully examining and comparing the results and acknowledging the enhancement. I appreciate the detailed description of your workflow; and I know exactly what you had to do in order to detect and correct the point "errors".

              Your suggestion is thoughtful. Thank you for bringing it up. Additional computation would improve the INSIDE point location - giving more weight to centroid location and honoring a specified distance from boundary. You may submit an enhancement request. I will definitely look into the algorithm and discuss the way to further optimize the result.

              Thanks for the feedback.

              Dan (Geoprocessing)
              • Re: Feature to Point questions
                goh_raj
                Hi Dan,

                Thank you for being receptive to my ideas. :D  Yes, I will go ahead and file an enhancement request.

                Thank you again for all of your help and support!
                • Re: Feature to Point questions
                  aysanaa
                  Does anyone know how I can create feature to point  in Arcgis 10.2 without extention?
                  Thanks
                  • Re: Feature to Point questions
                    rfairhur24
                    Does anyone know how I can create feature to point  in Arcgis 10.2 without extention?
                    Thanks


                    It does not require an extension, it requires that you have purchased an Advanced license.  Without the correct license level you can't use that tool.

                    About the best you can do within ArcMap without that license level is calculate the centroid XY coordinates of your features, export the data as a table, and then use the Make XY Event Layer to create a point layer from the coordinates.  This method won't achieve the same results as the Features to Points tool, but it may get you close enough to what you need.