Du Bois Visualization Challenge 04

Routes of the african slave trade – W.E.B Du Bois

Recreating the the data visualization of W.E.B Du Bois from the 1900 Paris Exposition using modern tools. See the challenge presentation.

This week, the data provided¹ don’t match the visualization. So we are free to be more creative…

We’ll try to show who profited from the slave trade by looking at the origin of the ships involved.

Setup

library(tidyverse)
library(tidygeocoder)
library(leaflet)

options(scipen = 100)

# compute mode (from https://stackoverflow.com/a/45216553)
stat_mode <- function(x, return_multiple = FALSE, na.rm = FALSE) {
  if(na.rm){
    x <- na.omit(x)
  }
  ux <- unique(x)
  freq <- tabulate(match(x, ux))
  mode_loc <- if(return_multiple) which(freq == max(freq)) else which.max(freq)
  return(ux[mode_loc])
}

Data

We impute the missing data and do some cleaning then we sum the slaves numbers by ships’ home port.

data_04 <- read_csv("routes.csv", na = "NA")

ports <- data_04 |> 
  group_by(ship_name) |> 
  mutate(port_origin = if_else(
           is.na(port_origin), 
           stat_mode(port_origin, na.rm = TRUE), 
           port_origin),
         port_geo = str_replace(port_origin,
           ", port unspecified|, colony unspecified|, location unspecified",
           ""),
         port_geo = case_match(port_geo,
           "Southeast Brazil"   ~ "Rio de Janeiro", 
           "Princes Island"     ~ "Sao Tome",
           "Para"               ~ "Belém", 
           "Lyme"               ~ "Lyme Regis",
           "Les Sables"         ~ "Les Sables d'Olonnes", 
           "Charlestown"        ~ "Boston",
           "Cabanas"            ~ "Cabañas, Cuba",
           "Camaret"            ~ "Camaret-sur-Mer", 
           "Goree"              ~ "Gorée", 
           "Saint-Louis"        ~ "Saint-Louis, Saint-Louis, Sénégal", 
           "Montrose"           ~ "Montrose, Scotland", 
           "Salem"              ~ "Salem, Massachussets", 
           "Stockton"           ~ "Newcastle", 
           "Cardenas"           ~ "Cárdenas, Cuba", 
           "Newbury"            ~ "Newbury, Massachussets", 
           "Norfolk"            ~ "Norfolk, Virginia",
           "Portuguese Guinea"  ~ "Guinea-Bissau",
           "Ilho do Fayal"      ~ "Azores",
           "Lancaster"          ~ "Lancaster, UK",
           "British Americas"   ~ "New England",
           "Warren"             ~ "Warren, Rhode Island",
           "St. Thomas"         ~ "United States Virgin Islands",
           "Danish West Indies" ~ "United States Virgin Islands",
           "Mediterranean coast (France)"         ~ "Marseille",
           "Sao Tome or Princes Island"           ~ "Sao Tome",
           "Spanish Caribbean, unspecified"       ~ "Havana",
           "Spanish Circum-Caribbean,unspecified" ~ "Havana",
           "Catuamo and Maria Farinha"            ~ "Pernambuco",
           .default = port_geo), 
         n_slaves_arrived = if_else(is.na(n_slaves_arrived), 
                                    round(median(n_slaves_arrived, na.rm = TRUE)),
                                    n_slaves_arrived)) |> 
  group_by(port_geo) |> 
  summarise(n_slaves = sum(n_slaves_arrived, na.rm = TRUE)) |> 
  filter(n_slaves > 0) |> 
  drop_na(port_geo) |> 
  arrange(desc(n_slaves))

Then we geocode:

ports_geo <- ports |> 
  geocode(port_geo, method = "osm")

Warning

Although I did clean the data, some errors in geocoding may persist.

Map

ports_geo |> 
  leaflet() |> 
  addTiles() |> 
  addCircleMarkers(radius = ~ sqrt(n_slaves) / 50,
                   popup = ~ paste0("<strong>", port_geo, "</strong><br />",
                                    format(n_slaves, big.mark = ","), 
                                    " slaves shipped"))

Figure 1: Home ports of the ships involved in the slave trade

Footnotes

1. The dataset seems to come from https://www.slavevoyages.org/voyage/database.