The same map under several classifications

Small multiples of one choropleth, drawn once per classification method, plus the break table and the count of countries in each class. The point is that the choice is consequential and usually unexamined: Brewer & Pickle (2002) found quantiles among the best methods for general choropleth reading and natural breaks (Jenks) below 70% as accurate, which is the reverse of the common GIS default.

Usage

classify_compare(
  data,
  value,
  methods = c("quantile", "jenks", "fisher", "headtails", "equal", "pretty"),
  n_bins = 5,
  ncol = NULL,
  ...,
  n = deprecated()
)

Arguments

data: A map-ready frame (polygon or sf).
value: The value column (unquoted).
methods: Classification styles to compare. Any of "quantile", "jenks", "fisher", "headtails", "equal", "pretty" and "sd", with the breaks world_map() would draw for each. "jenks" and "fisher" need the optional classInt; without it they fall back to quantile breaks with a warning.
n_bins: Number of classes (default 5), as the map verbs call it; "headtails" and "pretty" choose their own.
ncol: Number of facet columns.
...: Passed to world_map().
n: Use n_bins.

Value

A faceted ggplot object, with the per-method break and class-count table attached as the "countryatlas_classification" attribute (and readable with map_provenance()). Each method's rows also carry three measures of fit: gvf, the goodness of variance fit, and tai, the tabular accuracy index (both Jenks & Caspall 1971; 1 is a perfect fit), and max_class_share, the share of countries in the fullest class.

No number picks a classification

GVF and TAI measure how homogeneous the classes are, which is what natural breaks optimise; they rate a map that puts most countries in one class highly, and they say nothing about how well readers read the map. They are reported to inform the choice, never to rank the methods: Brewer & Pickle's finding that quantiles read best stands beside them, and max_class_share flags the map whose fullest class swallows the rest.

References

Brewer, C. A. & Pickle, L. (2002). Evaluation of methods for classifying epidemiological data on choropleth maps in series. Annals of the Association of American Geographers 92(4), 662-681. doi:10.1111/1467-8306.00310

Jenks, G. F. & Caspall, F. C. (1971). Error on choroplethic maps: definition, measurement, reduction. Annals of the Association of American Geographers 61(2), 217-244. doi:10.1111/j.1467-8306.1971.tb00779.x

Jiang, B. (2013). Head/tail breaks: a new classification scheme for data with a heavy-tailed distribution. The Professional Geographer 65(3), 482-494. doi:10.1080/00330124.2012.700499

Examples

# \donttest{
snap <- countryatlas::world_snapshot$countries
cmp <- attach_geometry(snap, geometry = "polygon") |>
  classify_compare(gdp_per_capita)
attr(cmp, "countryatlas_classification")
#> # A tibble: 30 × 7
#>    method   class              n   share   gvf   tai max_class_share
#>    <chr>    <chr>          <int>   <dbl> <dbl> <dbl>           <dbl>
#>  1 quantile 269 to 1.68K      40 0.201   0.626 0.672           0.201
#>  2 quantile 1.68K to 4.65K    40 0.201   0.626 0.672           0.201
#>  3 quantile 4.65K to 10.3K    39 0.196   0.626 0.672           0.201
#>  4 quantile 10.3K to 30.1K    40 0.201   0.626 0.672           0.201
#>  5 quantile 30.1K to 247K     40 0.201   0.626 0.672           0.201
#>  6 jenks    269 to 13.1K     129 0.648   0.958 0.756           0.648
#>  7 jenks    13.1K to 34.8K    38 0.191   0.958 0.756           0.648
#>  8 jenks    34.8K to 68.1K    25 0.126   0.958 0.756           0.648
#>  9 jenks    68.1K to 117K      6 0.0302  0.958 0.756           0.648
#> 10 jenks    117K to 247K       1 0.00503 0.958 0.756           0.648
#> # ℹ 20 more rows
# }