mafa

library(mafa)
library(ggplot2)

Compute orientations

Load data of parcels of the city of Rouen in 1827 and extract segments (lines) from polygons.

# transform polygons as segments (lines)
linerouen <- mafa_geom_to_segment(sfobject = rouen_1827, to = 'LINESTRING')

morphal_geom_to_segment() can be applied on LINESTRING, e.g. in the case of streets modeled as lines.

Compute orientations of segments with East looking and perpendicular parameters.

orientationsest <- mafa_segment_orientation(sfsegments = linerouen, looking = 'E', perpendicular = TRUE)

orientationsest |>
  ggplot() +
  geom_sf(aes(color = orientation)) +
  scale_color_viridis_c() +
  theme_bw() +
  theme(axis.ticks = element_blank(), axis.text = element_blank(), panel.grid = element_blank())

Compute morphological indices of polygons

Example with distance to minimal bounding rectangle (DSR):

rouen_1827 |>
  mafa_dsr() |>
  ggplot() +
  geom_sf(aes(fill = dsr), color = 'grey90', linewidth = 0.05) +
  scale_fill_viridis_c(direction = -1) +
  theme_bw() +
  theme(axis.ticks = element_blank(), axis.text = element_blank(), panel.grid = element_blank())

Compute clusters from indices

Compute indices.

rouen_with_indices <- rouen_1827 |>
  mafa_circularity() |>
  mafa_dsc() |>
  mafa_dsr()

rouen_with_indices <- rouen_with_indices[,2:5] # without rowid
rouen_with_indices
#> Simple feature collection with 10240 features and 3 fields
#> Geometry type: POLYGON
#> Dimension:     XY
#> Bounding box:  xmin: 560973.9 ymin: 6927999 xmax: 563141.3 ymax: 6929327
#> Projected CRS: RGF93 v1 / Lambert-93
#> # A tibble: 10,240 × 4
#>                                           geometry miller_index       dsc    dsr
#>                                      <POLYGON [m]>        <dbl>     <dbl>  <dbl>
#>  1 ((561754.2 6928994, 561759.5 6928992, 561756.2…        0.690  0        0.174 
#>  2 ((561779.4 6929002, 561773.6 6928982, 561764.7…        0.689 -3.19e-12 0.0625
#>  3 ((561732.2 6928970, 561735.3 6928967, 561741.4…        0.247  2.30e- 1 0.405 
#>  4 ((561715.6 6928974, 561708.7 6928977, 561705.1…        0.180  5.55e- 1 0.717 
#>  5 ((561605.9 6928902, 561603.4 6928896, 561592.8…        0.719  0        0.0265
#>  6 ((561604.7 6928915, 561607.3 6928914, 561614 6…        0.490  3.56e- 3 0.0603
#>  7 ((561610.9 6928904, 561611.7 6928904, 561610 6…        0.493  2.79e- 4 0.0623
#>  8 ((561621 6928889, 561612.5 6928867, 561608.2 6…        0.441  1.20e- 2 0.0711
#>  9 ((561627.2 6928869, 561623 6928871, 561619.3 6…        0.773  6.85e- 3 0.0271
#> 10 ((561628.3 6928906, 561621.3 6928890, 561616.4…        0.680  2.91e- 2 0.122 
#> # ℹ 10,230 more rows

Create HCA clustering and plot result:

clusterrouen <- mafa_clustering(sf = rouen_with_indices, pca_center = TRUE, pca_scale = TRUE, hca_method = 'ward.D2')

plot(clusterrouen, labels = FALSE, hang = 0)

Clusters

Cartography of clusters:

mafa_clusters(sf = rouen_with_indices, clustering = clusterrouen, cutting = 6) |>
  ggplot() +
  geom_sf(aes(fill = cluster), color = 'white', linewidth = 0.02) +
  ggthemes::scale_fill_colorblind() +
  theme_bw() +
  theme(axis.ticks = element_blank(), axis.text = element_blank(), panel.grid = element_blank())

Summary of clusters with mean of center-scale values of variables by cluster.

mafa_clusters_mean(sf = rouen_with_indices, clustering = clusterrouen, cutting = 6)
#> # A tibble: 6 × 4
#>   cluster   miller_index    dsc     dsr
#>   <fct>            <dbl>  <dbl>   <dbl>
#> 1 cluster 1        0.131  0.118  0.367 
#> 2 cluster 2       -0.152 -0.551 -0.632 
#> 3 cluster 3       -1.14   1.38   1.38  
#> 4 cluster 4       -2.20   3.54   2.92  
#> 5 cluster 5        0.991 -0.593 -0.688 
#> 6 cluster 6       -1.95  -0.359  0.0322