I realize that ESRI is calling their implementation of Fisher-Jenks proprietary. Given the performance with mid-sized datasets, either a significant optimization or a sampling method are being employed. I have found this link that used to point to pseudo code describing the implementation specifics.


Does anyone either 1) have published pseudocode information that they happened to archive for a rainy day, or b) have any information on the implementation details. In the latter case, not need to describe Fisher-Jenks as that algorithm is well documented. I am interested specifically in the ESRI implementation.


1 Answer 1


By default, ArcGIS samples for classification by taking the first 10,000 records. This can be changed in the classification dialog by increasing the number of records used.

For information on the implementation, I'd recommend seeing this mapping center post (and read Charlie Frye's comments since he worked on the original implementation)

  • Thanks for the info. The link you posted is the one I originally linked to as well. The poster specifically states that this is an ESRI version of Fisher-Jenks. I wonder if any additional information is available - the hand waving in the linked info does not speak to the specific alterations.
    – Jay Laura
    Commented Jan 27, 2014 at 4:15
  • It's not hand waving, particular focus areas are documented in the link. Anything deeper is difficult to do without releasing source, which isn't going to happen. Commented Jan 28, 2014 at 7:00
  • 'You can find the pseudocode here: resources.arcgis.com/content/kbase?fa=articleShow&d=26442' Any insight here? The implementation issues that the post documents exist across variance based classification methods. They are not specific to Fisher-Jenks, nor is the info from Charlie more insightful that going to the original papers that document the algorithm. I appreciate the link, but it offers no more information than either the original work or a few hours spent implementing the algorithm.
    – Jay Laura
    Commented Jan 28, 2014 at 14:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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