« Going overboard with simplicity | Main | The index of an index is confusion »



Don't the white areas take care of the case where one choice does not dominate utterly? It's what I'd do.


Derek: those are two different things. Think of the original data as a vector at every location. What Katz did is to pick out the maximum element of each vector and then take the average of these max-elements spatially. White areas on these maps would indicate that there is large dispersion in the max-elements in those spatial locations.


A possibility is to use an additional category of "No clear choice" for when the most common is not say 20% greater than the next. This is just one of several questions that are still there. There are probably other ways of dealing with the paucity of data in some areas and with dealing with the boundaries. Many of them are problems which are shared with mapping problems in geospatial mapping and ecology. I went to a talk a few years ago on geospatial and it was surprising how only recently they had developed models for a lot of their mapping, and still had much to do.


This is kernel density estimation, right?

The comments to this entry are closed.


Link to Principal Analytics Prep

See our curriculum, instructors. Apply.
Kaiser Fung. Marketing analytics and data visualization expert. Author and Speaker. See my website. Follow my Twitter.

Book Blog

Link to junkcharts

Graphics design by Amanda Lee

The Read

Good Books

Keep in Touch

follow me on Twitter