tmap basics: facets

About the data

Three spatial data objects of the Netherlands are contained in tmap: NLD_prov, NLD_muni, and NLD_dist, at respectively province, municipality and district level. The columns are the following data variables plus an additional geometry column which contains the geometries (see sf package):

names(NLD_muni)
#>  [1] "code"               "name"               "province"          
#>  [4] "area"               "urbanity"           "population"        
#>  [7] "pop_0_14"           "pop_15_24"          "pop_25_44"         
#> [10] "pop_45_64"          "pop_65plus"         "dwelling_total"    
#> [13] "dwelling_value"     "dwelling_ownership" "employment_rate"   
#> [16] "income_low"         "income_high"        "edu_appl_sci"      
#> [19] "geometry"

Facets or small multiples are small visualizations (maps), one for each group. The groups can be defined using a categorical data variable. In the following plot, a facet is created for each province (a data variable in NLD_muni):

tm_shape(NLD_muni) +
  tm_polygons("employment_rate") +
tm_facets("province")

In this example, there is one facet variable. For these kind of plots, and the facets are either wrapped or stacked, with tm_facets_wrap() or tm_facets_stack(). The general purpose function tm_facets() determines the faceting type, which is in this example wrap.

Groups can also be defined by specifying multiple data variable names for one visual variable:

tm_shape(NLD_muni) +
  tm_polygons(
    fill = c("employment_rate", "dwelling_value", "income_high"),
    fill.scale = 
      list(tm_scale_intervals(values = "-brewer.reds"),
         tm_scale_intervals(values = "brewer.purples"),
         tm_scale_intervals(values = "brewer.blues")))

Because different data variables are used, tmap uses (by default) multiple scales, one for each variable. These can be changed by putting the tm_scale() functions in a list, as shown above. In case the same scale should be used, fill.free should be set to FALSE:

tm_shape(NLD_muni) +
  tm_polygons(
    fill = c("pop_0_14", "pop_25_44", "pop_65plus"),
    fill.legend = tm_legend("Percentage"),
    fill.free = FALSE)

Facets can also be defined for two or three dimensions, obtained via tm_facets_grid() These three dimensions correspond to rows, columns, and pages respectively. The most common use case is two just dimensions, rows and columns. In that case, tm_facets_grid() is used.

tm_shape(NLD_muni) +
  tm_polygons(
    fill = c("employment_rate", "dwelling_value", "income_high"),
    fill.scale = 
      list(tm_scale_intervals(values = "-brewer.reds"),
         tm_scale_intervals(values = "brewer.purples"),
         tm_scale_intervals(values = "brewer.blues"))) +
tm_facets_grid(columns = "province")

(tm = tm_shape(NLD_muni) +
  tm_polygons(
    fill = c("pop_0_14", "pop_15_24", "pop_25_44", "pop_45_64", "pop_65plus"),
    fill.legend = tm_legend("Percentage"),
    fill.free = FALSE) +
tm_facets_pagewise())

will produce 5 plots, one for each age class (defined using the fill argument).

This plot can be exported as animation via tmap_animation.

tmap_animation(tm, filename = "NLD_pop_age_class.gif")

About the data

Facets: one dimensional

Facets: two or three dimensions

Facets as ‘pages’ (use case: animations)