Odd Coloring with GeoRegionValuePlot

Question

---

Bug introduced in 10.0 and persisting through 10.3 or later

Bug behavior is version dependent

---

When I'm plotting data with GeoRegionValuePlot, I'm getting some strange coloring errors:

The grid included shows the top 10 countries with arable land per population. However, Australia is colored the as if it were one of the least arable.

I thought it may be due to units, so I tried messing around with TargetUnits to no luck.

I figure that it's due to the fact that the data is being stored as exponents (i.e., the US's arable land/population is 4.81068*10^-3, so that's why it would be colored as if it were higher), but when I made all data have the same exponent, that didn't change anything either.

Any ideas?

Code, for reference:

(* Utility functions *)
removeMissingData[ data_ ] := DeleteCases[ data, _?( ! FreeQ[ #, _Missing ] & ) ]

divideStatsData[ data1_, data2_ ] := ( Module[ { data }, (* Module makes declared variables local *)
   data = { #, CountryData[ #, data1 ], CountryData[ #, data2 ] } & /@ 
     CountryData[];
   (* Missing data / missing data = 1 *)
   data = removeMissingData[ data ];
   { #[[1]], #[[2]] / #[[3]] } & /@ data
   ] 
  )

extractData[ data_ ] := QuantityMagnitude @ Flatten[ Transpose[ data ][[ 2 ]] ]

drawMapWithTooltips[ data_ ] := ( Module[ { rawData, min, max },
   rawData = extractData[ data ];
   { min, max } = { Min[ rawData ], Max[ rawData ] };
   GeoRegionValuePlot[ 
    data,  
    PlotRange -> { min, max },
    GeoLabels -> ( 
      Tooltip[ 
        #1, (* Location *)
        Extract[ 
         Extract[  (* First extract removes ugly braces, second actually gets information *)
          data, 
          Position[ data, #2, 2 ] + { { 0, 1 } } (* Search through table by country (#2) *)
          ], 1 ]
        ] &
      )
    ]
   ] 
  )

sortFromGreatestToLeast[ data_ ] := Sort[ data, #1[[2]] > #2[[2]] &]

(* Data *)    
arableLandPerPopulation = divideStatsData[ "ArableLandArea", "Population" ];
drawMapWithTooltips[ arableLandPerPopulation ]
greatestToLeast = sortFromGreatestToLeast[ arableLandPerPopulation ];
Grid[ Take[ greatestToLeast, 10 ], Frame -> All ]
Grid[ Take[ Reverse[ greatestToLeast ], 10 ], Frame -> All ]

Or if you don't want to load all that data:

arableLandPerPopulation = {{Entity["Country", "Australia"], 
  Quantity[0.0200457, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Kazakhstan"], 
  Quantity[0.0136973, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Canada"], 
  Quantity[0.0130036, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Argentina"], 
  Quantity[0.00939364, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Niger"], 
  Quantity[0.00890364, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Russia"], 
  Quantity[0.00840941, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Ukraine"], 
  Quantity[0.00731436, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Lithuania"], 
  Quantity[0.00692383, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Paraguay"], 
  Quantity[0.00637857, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Niue"], 
  Quantity[0.00620732, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Belarus"], 
  Quantity[0.0058311, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Guyana"], 
  Quantity[0.00551926, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Latvia"], 
  Quantity[0.0053108, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Moldova"], 
  Quantity[0.00522219, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "SaintHelena"], 
  Quantity[0.00521512, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Uruguay"], 
  Quantity[0.00514164, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "UnitedStates"], 
  Quantity[0.00481068, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Estonia"], 
  Quantity[0.00464188, ("Kilometers")^2/("People")]}, {Entity[
   "Country", "Serbia"], 
  Quantity[0.00458116, ("Kilometers")^2/("People")]}};
greatestToLeast = arableLandPerPopulation;

extractData[ data_ ] := QuantityMagnitude @ Flatten[ Transpose[ data ][[ 2 ]] ]
drawMapWithTooltips[ data_ ] := ( Module[ { rawData, min, max },
   rawData = extractData[ data ];
   { min, max } = { Min[ rawData ], Max[ rawData ] };
   GeoRegionValuePlot[ 
    data,  
    PlotRange -> { min, max },
    GeoLabels -> ( 
      Tooltip[ 
        #1, (* Location *)
        Extract[ 
         Extract[  (* First extract removes ugly braces, second actually gets information *)
          data, 
          Position[ data, #2, 2 ] + { { 0, 1 } } (* Search through table by country (#2) *)
          ], 1 ]
        ] &
      )
    ]
   ] 
  )

drawMapWithTooltips[ arableLandPerPopulation ]
greatestToLeast = sortFromGreatestToLeast[ arableLandPerPopulation ];
Grid[ Take[ greatestToLeast, 10 ], Frame -> All ]
Grid[ Take[ Reverse[ greatestToLeast ], 10 ], Frame -> All ]

Answer 1

Not a bug - it's a feature!!

First let's import the data in as quick a fashion as possible,

arableLandPerPopulation=<<"https://gist.githubusercontent.com/jasondbiggs/9e915145a2d4cfa34119fa4d0e535ed2/raw/75387a17c21ee7ff9d7f21e528ca2f971e24209b/gistfile1.txt";

Now look at the GeoRegionValuePlot with and without the PlotRange specified

GeoRegionValuePlot[arableLandPerPopulation, 
   PlotRange -> #] & /@ {Automatic, {0, 0.02}}

Notice the odd coloring in the second plot, that doesn't match the data? You can reproduce that coloring exactly by just ordering the countries alphabetically:

countrylist = arableLandPerPopulation[[All, 1]];
GeoRegionValuePlot[Thread[countrylist -> Range@Length@countrylist]]

So apparently when you give GeoRegionValuePlot a PlotRange it doesn't like, it will default to coloring the countries in alphabetical order. This is in 10.4 - in 10.0 they countries are all colored with the lowest value. Now why didn't it like the specified PlotRange? Because the data given had units!

InputForm[arableLandPerPopulation[[1]]]
(* {Entity["Country", "Afghanistan"], Quantity[0.00218678, "Kilometers"^2/"People"]} *)

If we specify the PlotRange using units, then the problem disappears,

GeoRegionValuePlot[arableLandPerPopulation, 
 PlotRange -> Quantity[{0, .02}, "Kilometers"^2/"People"]]

Okay, you can say it's a bug because it doesn't spit out an error message when the plotted units are Quantity objects and the PlotRange are just numbers, but this is explainable behavior.

Edit - Still a bug!!!

Try to set the upper range of the PlotRange to anything but the max, and it has no effect

GeoRegionValuePlot[arableLandPerPopulation, 
 PlotRange -> Quantity[{0, .001}, "Kilometers"^2/"People"]]